This PDF has been traveling the email circuit for a couple weeks, but it seems to have been posted online (or at least I’ve finally found a link to it).
A Proposal for National Economic Recovery An Investment in Geospatial Information Infrastructure
Building a National GIS Jack Dangermond, ESRI Anne Hale Miglarese, Booz Allen Hamilton
Salvation will be delivered with SOA!
98 Comments
Hmm, what is the date on that PDF? Seems like a solution from the 1970s, not 2009.
I need to get me some of that sweet, sweet pork! I’m fine with ESRI proposing this to be honest, but BAH? They are part of the problem!
No one ran this through us first. Jack, you need to get clearence with Google before proposing technology to Barack.
Standardize zoning codes nationally? How many lifetimes will that take…and that just nailing down the politics. I’d love to see it, but man, getting some of them municipalities to agree would be a total coup. Then, deciphering the legalese of their zoning codes could take a decade or two.
This would be a lot more interesting if they had partnered with other geo technology companies, the OGC, OSGeo, or pretty much anyone other than a beltway bandit. You can also most see the emails – we’ll take care of the lobbying and you supply the tech then we’ll get a huge contract to implement it all.
“Existing modern GIS server technology…can provide enabling components for a national GIS immediately”
Someon needs to tell them they forgot to put “Arc” in front of GIS. On the upside this really makes the case for where a Federal CTO could be really useful.
What are the odds of a Federal CTO not being an ESRI fanboy?
Sean,
I seriously doubt anything substantial will come out of this. It’s very preliminary, but I do understand the first shot over the bow approach.
More people lining up at the trough. Bunch of pigs!!
Centralization of everything?? sigh
“”It’s not that I want to punish your success. I just want to make sure that everybody who is behind you, that they’ve got a chance at success, too… My attitude is that if the economy’s good for folks from the bottom up, it’s gonna be good for everybody. If you’ve got a plumbing business, you’re gonna be better off [...] if you’ve got a whole bunch of customers who can afford to hire you, and right now everybody’s so pinched that business is bad for everybody and I think when you spread the wealth around, it’s good for everybody.”
You see, Jack just wants to help (put) everyone behind him!
Great a $1.1 billion aid package for ESRI, how apt along with the other dinosaurs who have their caps in hand.
That plan looks like a plan for government based on government recommendations.
I do think, however, that the value of a National Map as described in the paper is an excellent goal. Take a look at ParcelAtlas.com for a look at one of our projects that is contributing toward that goal.
I’m hoping we get a piece of that $200 million they’ve carved out for parcel data!
Not so. Business is booming for modern, cost-effective GIS vendors as buyers flee from ripoff legacy vendors. Layoffs might be a big topic at Pitney Bowes and ESRI, but modern GIS companies are rolling in cash and don’t need no stinkin’ bailout.
The recession is forcing more and more legacy GIS users to seriously consider for the first time more modern alternatives, and they like what they see. In good times ESRI could get away with a lot – In bad times, GIS buyers realize that stupid procurement can cost them their jobs. The economic collapse makes it a lot harder to spend $30,000 on living fossil GIS technology to get less than what $500 spent on modern GIS technology buys. So yeah, you bet ESRI is trying to get a bailout like their life depends on it. It probably does.
Of course, nobody’s going to give you a bailout. It’s going to be up to you to make sure you keep your job while ESRI’s off trying to pick Uncle Sam’s pocket.
See my comments in http://www.directionsmag.com/article.php?article_id=2973 on practical measures to help you keep your job during this recession.
ESRI, layoffs? tell us more ohhh wise GIS industry insiderman!
Wrong again Dimitri. Go to http://careers.esri.com/ and see if ESRI looks like they are laying off.
Ummm, Adam, I guess you don’t read the financial news. By the logic you are applying, because Microsoft’s career websites are still advertising many positions the financial news reported yesterday and today that Microsoft is preparing significant layoffs for the first time in its history could not be true. Somehow, I don’t think the Wall Street Journal will retract their story based on your logic.
Not to put too fine a point on it, one reason dinosaurs don’t do well in recessions is that they can’t adapt to the changed circumstances. So they continue to waste a lot of effort in HR hiring people in one area while they are laying off more folks in a different area.
In the context of that, let’s consider this “wrong again” comment you made. Put up, or shut up, Adam, as I’ve not been wrong about any of this. I predicted that ESRI would seek a bailout and indeed they have. Yep. Check. You don’t think I was guessing about that, do you?
In the current thread I wrote that layoffs are a topic of conversation, not that they have happened already. I trust everyone here understands the difference given my habitual precision of language, and that everyone also realizes that actual layoffs are not so easy to hide.
Adam, seriously, are you really willing to state for the record in this forum that you are completely certain that layoffs are not a topic of discussion at ESRI? Don’t be shy: here’s your chance to get archived forever as someone who had no knowledge of anything like that at this time. Since you disagree with my views, let’s get your view on the record.
To help break the ice, I’ll be happy to go on the record, and I can do so without breaking any confidences.
I personally don’t think there’s anything wrong with Microsoft, or AutoDesk, or Pitney Bowes or ESRI having internal conversations about layoffs given the economic stresses they face. That is the intelligent, commendable thing for them to do and they’d be failing in their duties as managers if they did not discuss the matter with the seriousness their changed financial fortunes demand.
Going from managers down to the employees they manage, I suppose just about everyone here is savvy enough to know that anyone with a shred of self preservation instinct within ESRI has no doubt considered the matter as well. They’d have to be stupid not to, and they’re not stupid. The ESRI bailout plan did not just materialize out of the sky because they think their finances are the best they could possibly be.
My point is simply that they are facing those undesireable trends, and, further, I’d expand that point by stating that they are reacting to those trends not by reforming their product lines, or by modernizing their products or by offering better price/performance, but by trying to deal with the situation by picking Uncle Sam’s pocket.
Adam inadvertently fanned that fire with the link he provided: one reason dinosaurs get killed off by recessions is that they are simply unable to change their behavior to adapt. It’s classic dinosaur brain behavior to be unable to contain their cost structures within the new constraints of a recession. It’s the classic dinosaur move to be hiring out of inertia even as you are firing elsewhere and worried about just how many layoffs you will have to do to survive.
Let me close this by making a crystal-clear prediction: if ESRI does not get their bailout and if the recession keeps going as it has or deepens, then ESRI will indeed be laying off some personnel before the end of 2009. Let’s see who called that correctly when the end of 2009 rolls around, right on schedule towards the end of December. Soon enough!
Just noticed another one of Dimitri’s gems: “…you bet ESRI is trying to get a bailout like their life depends on it. It probably does”.
Cash rich ESRI’s life doesn’t depend on a ‘bailout’ to survive. C’mon… that’s a stretch. This proposal is a way for them to continue to lobby at the upper level and hopefully get lots of new and lucrative Federal level jobs that other smaller GI companies could never take on. I don’t see this as a rescue from financial difficulties – more of trying to seize on a lucrative opportunity.
I’m not saying I agree with it, but don’t think that ESRI needs this to survive.
“Cash rich ESRI’s life doesn’t depend on a ‘bailout’ to survive. C’mon… that’s a stretch. “
Unless you know ESRI’s internal cash position and their burn rate you don’t know. ESRI has huge costs, far greater than their much leaner competitors.
But suppose I grant you your thesis, that ESRI feels cash rich. If that’s the case, that still doesn’t mean that they don’t realize that they need a bailout to survive and would prefer to use Uncle Sam’s money than burning through their own. They’re not stupid. They also obviously have no moral qualms about redirecting money taken from struggling taxpayers into corporate welfare for bloated Federal contractors.
Let’s talk about cash for a bit, because the trappings of past success can end up being an obstacle to future success in the case of dinosaurs.
A strong internal cash position is exactly one of the factors that can often prevent dinosaurs from acquiring the agility they need to deal with changed circumstances. That is precisely a factor that helped kill off Wang, killed off DEC and now seems to be immobilizing Microsoft in the face of upstart competition from the likes of Google. It’s the old “What do I have to worry about? I have billions in the bank.” I have worked inside some of the largest companies and I can assure you it is almost impossible for people to think that a day of reckoning will come when billions of dollars are floating around to insulate them from reality.
You have to look at their strategic position: if they are in a disadvantageous position and aren’t able to change that, well, then there comes a time when the cash declines down to zero.
In ESRI’s case they have retreated from fierce competition in the mainstream desktop, abdicating that market to the degree they don’t even exist in it. Instead, they focussed on desktops in the enterprise. Now they are retreating from that to focus on a more controlled IT sell by trying to move customers to server-side architectures. Their bailout proposal is a deliberate choice to leverage their government business, which is their last bastion against competition. These are all structural/political moves designed to protect a past technological status quo and it is not innovation.
What you don’t see from ESRI is the sort of technological ferment that characterizes a living, muscular company that will thrive long term in technology businesses. It’s as if Sony, after being the king of cassette players and Walkmen would end up deciding the way to deal with the rise of MP3 players is to lobby the government to standarize on cassette player Walkmen to equip every Census taker for their entertainment. You might think “Yeah, if they get enough Congresspeople to sign off on that it will happen” but you’re not going to take that as a sign that the strategy would be a winner competing against the iPod.
When ESRI charges ten to forty times as much to deliver less capable and less reliable technology than more modern vendors it is only a matter of time before they get extinguished if they don’t remedy that fundamental price/performance disparity. It doesn’t matter how much cash they have if they don’t use it to remedy the fundamental problem. Where we see that cash being used is on lobbyists, not more effective engineering.
I don’t mean to suggest that ESRI are idiots for doing the above either. They’re not. They are in a tough box where it is extremely difficult for them to modernize without messing up their income. After all, just how would ESRI convince people to buy their current product line if they had a modern alternative that did more at a tenth of the price? It’s not so easy to make that transition.
Lay offs at ESRI? Then why is there an office construction boom under way at ESRI HQ in Redlands? One big building just occupied last fall, another big one including an auditorium nearly complete, and a major expansion of the parking lot. Not to say that ESRI isn’t aware of the competitive and economic threats to its business, but as of now, things look under control. Sales up, profits up, recruiting like crazy, all this 23 months into the recession. The real message here is one of a financially conservative leader who didn’t cash out and become a publicly traded Wall Street whipping boy. Greedy stockholders have crushed more competitors for ESRI than any release of ArcInfo ever did (remember MapInfo?). Dangermond has been saving for this rainy day for a long time. The “Bailout”? meh, that’s just Dangermond doing what he always does: promoting GIS.
Interesting that you won’t commit yourself, “ESRI Insider.” If you disagree with me, let’s hear you say it. Just say those magic words…
a) “No, I’m sure layoffs have never been discussed at ESRI.” and…
b) “Yep. I’m sure that if ESRI doesn’t get their bailout and if the recession continues or deepens there will be no layoffs at ESRI.”
Go ahead, ESRI Insider… well? Do you feel lucky?
Can’t resist: “all this 23 months into the recession.” Really? I guess you haven’t heard of the economic collapse towards the end of 2008 as being something distinctly different from the high times in 2007.
…and…
All this saving for a rainy day stuff is fine and good and commendable, but Wall Street and public companies have their strengths and the well-managed public companies have huge rainy day funds as well (Microsoft’s coming to mind).
If the owners of ESRI are happy to burn up their wealth putting off a day of reckoning, that’s fine. History shows that you can’t make up a huge technology and price/performance disadvantage nor an absolute pricing disadvantage that way in high tech businesses – you just burn up the cash to no good end if you are unable to fundamentally change the way you develop technology or do business. That’s why the tech landscape is littered with so many corporate palaces left empty soon after they were built. If you are close enough to do so, go visit Silicon Valley for a tour of all the empty shells, or if you are on the East Coast pay a visit to all the Wang and DEC relics.
Can ESRI reform itself to produce modern technology and sell it at modern, competitive levels of price performance? If they’re chasing after bailouts to allow them to continue the fat headed ways of the past I don’t think so. If anything, everyone who would like to see ESRI reform should oppose this because the only thing that will cause them to really work hard at modern levels of price/performance instead of chasing the Federal $500 hammer and $2000 toilet seat thing is fear of a reckoning.
Dimitry I have heard nothing of layoffs at ESRI. I have heard no buzz of layoffs at ESRI. My team had a reasonable increase in revenue and profit in 2008. ESRI is still actively looking to hire talented people which we would not be doing if we were thinking of layoffs. ESRI has never had layoffs in its 30+ year history and this is not the first recession we have seen at the company. As much as I enjoy reading your rambling FUD postings all I can say is that Jack will (as usual) describe how ESRI is doing at the next UC.
Side note to recently laid off Intergraph and Autodesk employees: if you love GIS and want to work in the field at the largest and (cough) most stable company out there, send a resume to us.
Dimitry, I don’t know what my luck has to do with anything, but i know for a fact that layoffs have never been discussed at ESRI. How’s that? Do I feel lucky? As a a matter of fact, yes.
Obviously i can’t claim there would never be layoffs; nobody knows how bad this “Great Recession” will get. Forget ESRI, I’m actually more worried about the future of the Union. But to link this proposal by ESRI to its ultimate survival is absurd as it is naive.
I thank you for going on the record. We’ll see how it goes by the end of 2009 in this prognostication contest.
I disagree with
“But to link this proposal by ESRI to its ultimate survival is absurd as it is naive.”
My own view is that ESRI will probably get this bailout. It has been developed with a razor sharp understanding of how to feed at the public trough, the secret meetings in advance with Congresspeople have been done, and the bailout has been designed with plenty of bribes for the supporting cast that will help it become law.
But without the bailout you’d see ESRI under pressure for charging obscenely more while delivering less (it’s not like people don’t notice that sort of thing during grim recessions) and that is what starts the slide that makes it a matter of survival.
Oddly enough, it’s an open question which ultimately will be worse for ESRI, getting the bailout or not getting it. I think the consensus among my colleagues is that the bailout works as a time-shifter in that pushes off the day of reckoning but makes it more certain. I think most of the “modern GIS” crowd believes that a bailout will help take ESRI out of commercial circulation, perhaps not right away but with greater certainty than otherwise.
The reason is that a billion bucks is too much to ignore. Whatever comes out of that bailout, it will grab the attention of ESRI like nothing else and that’s where ESRI’s prime focus will be. The problem with that is that the demands of government usage unfettered by market realities has almost always in the history of technology resulted in products which are not effective competitors in open markets. Anyone remember Ada?
I think the consensus among my colleagues is that ESRI probably will get this bailout, they’ll be happy to have it, that will enable them to avoid layoffs throughout 2009 and even 2010, it will pay for some nice new buildings, and the result is that ESRI’s product family will drift even further away from what most people in the world who do GIS day in and day out need. That’s fine by me, because all of those folks are not going to want to give up with their desires to learn to live with the Ada of the GIS world, they’re going to go off and buy a product which is shaped for their needs both as regards functionality and as regards price.
I don’t think anyone is going to complain about government funding for geospatial technologies – it has the potential to be a very good investment. The question here is how do you get the most benefit for the citizenry out of that investment.
I’d argue that perpetuating a proprietary legacy technology at an even more massive scale by tacking on a few buzz words – open standards, SOA etc. etc. is not the answer. Instead invest in making government data open and freely available in standard formats so that companies can innovate around it. The Apps for Democracy project was a great example of this and it’s brainchild, Vivek Kundra, is one of the front runners for the Federal CTO post.
You can incentivize innovation with project grants that demand interoperability and open standards, then use the successes for deployment. There is going to still be a need to provide government systems for non-public data and the successes can compete for those contracts (ESRI, Google, MSVE, MApInfo you name it.). I’d have to think that competition and innovation will yield better results than a monlithic centralized system that will be behind schedule and over budget because the vendors and contractors make more money that way – no cynicism of course.
I’d think this is exactly the kind of thing the new administration would be looking to change.
Hmmm – if this email is legit, I think they’re waaaayyyy over simplifying the situation.
Bailout for ESRI et al eh, intersting. Considering what I’m hearing from some of our clients, it sounds like a plausible reason to put something like this out.
What e-mail? Can someone share the text of it?
Anyone know who this ESRI bailout package has been proposed to, if anyone?
I pledge allegiance to the National GIS Infrastructure and the ERSI Products for which it stands One Nation, under Versioning, With ArcInfo and AGS Advanced Enterprise for all (who can pay for it).
How about this, make a GIS program or even some kinda Google Earth mash up tool available to the public or people with basic GIS skills (at little or no cost). Create a grid, (or use an existing one) define data layers that are needed, assign a value for collection per grid (or layer) and let the taxpayers earn some cash collecting data. Most of what is needed can be heads up digitized from existing imagery anyway. Not a bailout, make it a workout!
If people are willing to scan tons of imagery looking for Steve Faucett, I”’bet they’d be willing to heads up digitize for some relief in this economy.
yeah, thanks, that approach worked out great for me.
Steve Fossett’s Plane Wreckage Found – by a Hiker
http://blog.entchev.com/2008/10/04/steve-fossets-plane-wreckage-found–by-a-hiker.aspx
OK, OK! I didn’t say the GIS community fond him. I said they were involved in the search. My point was that included in today’s GIS sphere is heads up digitizing which doesn’t take programming skills or a GISP, and could be done with minimal training by any high school grad. The money ESRI is after does not have to go to only a few priveledged companies such as Booz Allen.
OK … now I get it.
National System and Data Integration approximate cost $450 million
I imagine Jack was imagining something along the lines of: Four Years of 20 Administrators 200 Data Analysts 2,000 GIS Data Technicians 20,000 online users 10,000 Manifold licenses AtlasAlive On Demand
Obama’s Performance Overseer would demand nothing but!
20 administrators and 20,000 users? On a government project? I think you’ve got your pyramid the wrong way up!
Also on the web here http://www.gis.com/gisnation/. More clarification between this proposal and the National Map, NSDI, FGDC is needed.
N.B., The Obama transition team attended a meeting of the FGDC and has been asking about GLoB (government Geospatial Line of Business) and such.
If you are interested in a well researched argument with statistics that show that open records with respect to parcel datasets results in increased tax revenues, read my related blog post here:
Fueling National Economic Recovery through Investments in GIS
Okay – I’ll be the one to say it: Wouldn’t this money be better spent elsewhere? Don’t get me wrong – I’d like to see the government throw piles of cash into GIS as much as the next guy, but is now the right time to invest a billion+ into technologies that yield limited immediate benefits? I’m not saying that our government shouldn’t invest in GIS, I’m just wondering why we’re talking about making the investment right now. It just seems to me that anything we spend more than a billion on today should have at least a chance to help stabilize our shaky economy – immediately. I have to wonder if GIS fits this bill. With the economy as deep in the crapper as it currently is, I think we need to be looking more toward the short term. The backward motion needs to be stopped before the forward motion can begin.
Most of the money being spent will only increase the national debt and increase our future interest bills and taxes. If Congress were truly intent on stimulating the economy, they would pass an immediate, across the board tax cut rather than invest in welfare and political payoffs. Allowing people to keep and spend their own money is a sure thing (unless your purpose is to buy votes and retain your slot sucking from the public trough). Running that money throught the bureaucracy, with most to pay political cronies will create little benefit. Bail out spend thrift state and local governments? Bail out teachers and schools? Bail out universities? What a farce. Apparently, it has become un-American to have a little pride and self-respect. Welfare is where its at!
HEY does ANYBODY know who this proposal was written for or sent to? Is it just a set of talking points or is Jack actively trying to put this over with someone on the Obama team?
I think the following gotgeoint blog post might be interesting to many of you tracking this post:
http://www.gotgeoint.com/archives/coalition-of-geospatial-organizations-urges-congress-to-form-geospatial-subcommittee/
SOA has become a 4-letter word. (In the same way GIS should be per the 2nd article below)
http://apsblog.burtongroup.com/2009/01/soa-is-dead-long-live-services.html
http://apsblog.burtongroup.com/2009/01/soa-postmortem.html?ref=VortexMe&aff=VortexMe
http://www.eweek.com/c/a/Web-Services-Web-20-and-SOA/SOA-Wanted-Dead-or-Alive/
I could`ve written this “proposal” in my coffee break. Plus, come on, investing in ESRI´s technologies will NOT save any Economy whatsoever.
Are they serious?
Sadly, SAntiago is correct. JD should be commended for the great service he has brought to the GIS community. He has truly revolutionized the field.
That is what has made it so sad to see his name on this ham-handed proposal. It so much looks like a lifeline for ESRI, and most people see right through it.
It reminds me of watching Joe Namath in his last year with the Rams, or Johnny Unitas with the Chargers (or Favre if he returns to the Jets). It was sad seeing those guys go out the way they did.
Didn’t Favre return to the Jets? I’m not even a football fan and I know that.
I agree with you on everything else, though.
Yes. Favre played for the Jets last season (just about everyone knows that – whether they are a football fan or not). The question is, whether he will RETURN next season. He played very badly his last 6 games, so the question is, if he RETURNS next season, he will probably embarrass himself.
The rumours of ESRI’s demise are greatly exaggerated. This is not a bailout package for ESRI. The incoming Obama administration has committed itself to a stimulus package which targets specific areas of infrastructure for the nation. This has been done in the past by previous administrations. The Eisenhower administration championed the interstate highway system, as did the Kennedy administration the space program. In this century, it seems obvious that the targeted infrastructure should be communications and IT, including GIS.
The level of hatred that some supporters of open source GIS have for ESRI is unfortunate. ESRI, a for-profit company, has a impressive technology stack, but ESRI technology is not always the right solution to a problem. There are also a number of great open source GIS packages, but open source technology is not always the right solution to a problem. Through open standards and interoperability, open source, ESRI technology, and other technologies are just different parts in the bin, to be chosen as appropriate for each particular solution. Using ideology to guide solution choices can lead nowhere but to sub-optimal solutions, and ultimately does serve our customers and our community well.
Legacy vs modern GIS technology? Sounds definitive. Where is that line drawn?
Good question. I make that distinction a lot so I owe you a fair answer. There are many objective measures of what constitutes “modern” technology. Here are a few:
Computer hardware and software technology advances so rapidly that anything more than a couple of years old is usually far behind the capabilities, performance and reliability of current technology that can rightfully be called “modern” in any sense of the word. In many cases, stuff that is more than a few years old is literally worthless to the point you have to pay the dump to take your old gear. Can’t even give it away.
That’s one metric you can apply: if your GIS vendor does not embrace current technology they can be said to be a “legacy” vendor and not a “modern” GIS vendor. So how “current” is current before we say someone is not keeping up? Six months? A year? Two years?
I can hear ESRI salesmen howling now that we shouldn’t hold their feet to the fire to embrace new standards as rapidly and as effectively as, say, the video game industry. After all, everyone would no doubt agree that someone asking for a billion dollar bailout upon which the very economic future of the United States will be based should not, absolutely not, be as alert and as effective in implementing technology as some pimply-faced teenager fooling around in his bedroom playing Halo. Don’t want to make those technology expectations too high for a billion dollar bailout!
Let’s cut ESRI some slack and make the comparison easy, say, two years to more than five years back and look for objective comparisons there.
There are thousands of points of comparison between “legacy” and “modern” that could be cited, but given my lack of desire to write more than a few thousand words at a time let’s just look at the main technology standards that have emerged between the last five years or older and the last two years that might be especially relevant to GIS.
The biggest trend was the emergence of 64-bit processors and 64-bit Windows five years ago. For years now, just about every processor and computer system sold into professional and enterprise settings has been fully 64-bit hardware. If five years after the fact you can’t provide 64-bit GIS throughout your entire product line you are not a “modern” vendor. You’re a “legacy” vendor and you may as well go home, stick your head in the toilet and flush for being such a total loser.
The second biggest trend was how RAM became darned near free to plug in 8 GB or more. Lots of cheap RAM is the modern, easiest, least expensive way to get better performance and more reliability. If you can’t utilize that RAM (this is a corollary of not being able to do 64-bit operation), you ain’t modern, amigo and you may as well flush twice.
Next up is how processors have gone multicore. Intel and company ain’t stupid, and they only way they could get out from under the end of Moore’s Law was to go multicore. The Core 2 Duo has been around, what?, three or four years already? If you can’t do any multithreaded stuff, you ain’t modern either. This is when ESRI is reaching for that third flush.
Speaking of multicore and 64-bit, all that stuff has come in a setting within modern Windows releases like Vista and Windows Server 2008. If it takes you years to support Vista and you haven’t yet supported Server 2008, well, comrade, reach for that fourth flush because you’ve got yourself deep into “legacy” mode.
Coming up to the most recent big event, going back just over two years, the biggest news in modern computer science is how people are using massively multicore GPUs as general purpose compute engines for absurdly low costs, a hundred or two hundred dollars, and sometimes even for nothing because the system already has a supercomputer-capable GPU in it. Do that modern thing and you can provide supercomputer performance, literally accomplishing in seconds a task for which a legacy system like ESRI’s will require minutes. If a GIS does not use technology like CUDA, it is not a “modern” GIS.
[Readers interested in the technology of why Moore's Law came to an end and where multicore parallelism and massively parallel GPUs fit in should see the video at http://www.manifold.net/video/Supercomputer_GIS.wmv ]
Let me wrap this up with a few paragraphs of very specific interest to GIS.
Big trends in the last five years have been to greater integration. If your GIS vendor wants to sell you five different packages with five different object models to do desktop GIS, IMS, spatial DBMS, raster stuff, and development using objects, you are dealing with a legacy vendor. The modern way is to integrate all that so that you only need learn one object model, and so that any improvement in one area automatically benefits everything.
A second trend is that every big-time DBMS on the planet has introduced their own, direct, high performance spatial DBMS technology. If your GIS vendor wants to interpose a legacy (as in the literal sense of being left over from an earlier day) middleware techology that prevents you from using a modern spatial DBMS as the designers intended, there too you have a legacy GIS technology and not a modern GIS technology.
Speaking of DBMS, a modern GIS will incorporate superb DBMS capability within it, including a truly full-featured spatial SQL that can run locally to do analytics on the fly and create new types of data for display and use.
And then there’s the matter of the web being everywhere. If your GIS is not web-aware through and through, it’s just not a modern GIS technology. A modern GIS, for example, will include an integrated IMS so that it can publish within web applications, and it will be able to do so via HTML, WMS, WFS-T, image servers and tile serving. A modern GIS will also integrate client stuff, so that it will be able to automatically consume from image servers like Virtual Earth, Yahoo, TerraServer, Google Earth and NASA Worldwind, and it will be able to be a WMS client and so forth.
And last but not least a modern GIS will be sold in the modern way, not the legacy way. It will be priced the way modern software is, at $250 a copy or so, not tens of thousands of dollars for old-fashioned stuff. And it will be sold through the web in a perfectly transparent way as regards price. None of this “we’ll send some guy over in a suit to decide just how much you are going to have to pay us.”
The word “legacy” does indeed mean left over from an earlier day, so no doubt people who are intimately familiar with the various antique outcroppings of old technology within ESRI can cite more specifically why ESRI is well known as the embodiment of legacy GIS with specific, objective technical citations. For example, there is this odd persistance of limitations originally encountered in shapefiles that have persisted up through the ESRI technology of implementing shapefiles in database form into what is now the latest versions of SDE. I’ll leave that up to those who have an interest in software archaeology.
While I can’t agree with everything Dimitri says, he’s spot on with his comment about SDE when RDBMSs now natively support spatial datatypes and spatial SQL. In many respects, getting SDE is like getting a cable box for a cable ready TV.
Organizations need to take a close look to see if SDE is simply an extraneous piece to one’s technical reference model that doesn’t add alot of value given the total cost of ownership.
PS Jack stold my girlfriend in college and I’m still pissed!
Very funny, I suppose for those who know neither Jack nor me nor my girlfriends in college (I had to pause for a moment just then to let the irreplaceable warmth of that flood of memories wash over me… I sincerely hope everyone at least once in their lives has something so wonderful to remember).
Well, at first I thought it was a shame that anyone can pose as someone else and that a handful on this otherwise very elegant blog (kudos to James and the community he hosts…) don’t meet the high standard set by the majority.
But then I guess there’s no real need for any identity checks as those who really count here have far and away more than enough swiftness of mind and sharpness of eye to immediately recognize the turd in the punchbowl when it floats up (thanks go out to Norman Mailer for that memorable metaphor).
So Dimitri, in your opinion, are there any other ‘modern’ GIS technology companies out there besides Manifold?
Based on your definitions above, I’d like to see who else fits ‘modern’ in your mind.
Well, I didn’t mention any product names, not even the one you cite, explicitly to leave the matter open for individual readers to judge as they see best. If people look around and apply their own experience they’ll believe what they find all the more whether they agree or disagree with me.
Plus, not using product names puts focus on the criteria. I obviously feel strongly about those based on my personal experience and prejudices, but if anyone doesn’t like my suggestions for objective criteria that makes the difference between legacy and modern they can recommend their own. I’m curious to see what other people feel would be good criteria.
D – thx for those detailed posts. Your talking points are very poignant. Still, while I agree with what the demands of a discerning customer should be, it doesn’t seem to me that any vendor out there can really stand up to such scrutiny. If there was, ESRI would have sank years ago. I lol’d at the video game analogy – ESRI would be long gone or shucked into making kid’s games if that was it’s business.
There are cheaper GIS applications catching up to the Neanderthal, maybe if ESRI didn’t get the memo, drop your prices, & start modernizing.
Dmitri usually isn’t far off the mark but I will have to point out that he is wrong in one instance. ArcServer does use multiple processors/cores. You just have to pay for them. Our organization just received a couple of ArcServer licenses. (Enterprise, SDE, etc)
Each was in the range of (pause for affect) $100,000! He may have been referring to the desktop product and then he would of course be 100% correct.
I am just pissed at this point that it cost you to leverage current hardware standards. I have been a ESRI fan for the past few years but am getting sick and tired of having to put up with this sh*t. I still use ESRI products but am trying to find the best way to slowly migrate from the strangle hold that it has on my community. (Federal Government)
am trying to find the best way to slowly migrate from the strangle hold that it has on my community
the best way is to start learning how to use Manifold. Don’t be too aggressive, though. Just get a copy as an “experiment”. Learn how to use it, ask questions at the Manifold forum, and then test out some ideas with it and see how it goes.
It will take time, but slowly demonstrate your application to members of your staff over time.
I’ve been doing just that as a new entrant to GIS who can’t afford his own ESRI stack. It’s kind of slow going for this Geo-monkey wannabe, not least because of Manifold’s somewhat eccentric conceptual approach, although I really like the easy access to the RDBMS, among many other data formats, and the overall functionality seems very impressive for the price. But having read Dmitri’s passionate and informative posts here, I’m starting to suspect who’s responsible for all those barbed comments about “legacy GIS” vs “modern GIS” in the Manifold user tutorials!
Anyway, hope the US govt invests wisely if they’re going to spend all this money on GIS, rather than simply funnelling all the cash straight into the pockets of a few bloated vendors. Good luck with the GIS pork barrels, guys!
“He may have been referring to the desktop product”
Yes, that was my intent and I apologize for not being more clear.
This really deserves its own thread, but part of the economic challenges ESRI faces are rooted in the technology decision ESRI has made to emphasize server-side work over the desktop.
You can see that in how their Server products are moving forward into 64-bit and multicore, at least a little, while the desktop products aren’t being modernized at all. I would respectfully suggest that even in their server line, even with the commendable progress they are making there, despite the modernization with any 64-bit or multicore operation in the server line, at the end of the day as far as most real-world GIS operations are concerned this focus on server-side is deeply legacy stuff because it is saying you should go back to a timesharing model that was discredited many years ago when people realized that distributed processing on desktops was far more performant.
“Server side” is just time sharing: it is many clients playing “dumb terminal” to execute an application on a central server computer. So what if that central server has two, four or eight cores? If you have a hundred people connected to it it’s not going to be as much processing capacity as represented by the hundred or two hundred processors (most professional computers are dual core these days) on the desktops of those hundred users.
No way is one hundred users sharing one server going to be as fast as one hundred processors working all for a single user. That’s the key truth that killed off time-sharing in favor of distributed computing on the desktop.
I can understand that ESRI’s economic interests are served by centralizing applications onto a server. If you have a national accounts sales force you have to keep employed it is more effective to have them calling on the top IT guys, so that if they sell one centralized application at the top of the pyramid then all the unwashed users just have to fall into line and salute. It works for selling centralized applications to Federal agencies, as well. It is very smart of ESRI to use their sales force in this way and the move to pushing server-side solutions works well in their business model for ESRI’s economic benefit.
But it does not account for progress in modern computing technology nor does it serve the economic or day-in, day-out convenience of everyday GIS users.
When you do GIS for eight hours a day you want that GIS system to be a slave to your convenience, not the other way around. It should be fast, it should be convenient and it should be darned near as cost free as possible. The way to load the most performance for the buck is to take advantage of those processor cores that are spinning around at billions of cycles per second in your computer just waiting for a chance to do something productive for you. It’s not to blow them off to spend endless seconds waiting for some fraction of some weaker core at the tail end of a much slower Internet connection to wake up and loan a millisecond or two of work on your behalf.
I’m writing this on a stupid old Thinkpad notebook computer that for all its simplicity and age has 4 GB of RAM, two main processor cores and 32 stream processors with nothing else to do in their lives but to burn epic billions or trillions of cycles on my behalf. The sheer computational power of those 32 stream processors, packaged within a two year old NVIDIA FX quadro mobile GPU, would probably eat alive whatever is the hottest dual or quad socket Xeon server that you’ll find in any centralized server rack. All you have to do is run software that takes advantage of it. If you have a Thinkpad from Lenovo or any of the other very many laptop computers with a reasonably recent NVIDIA GPU in it, well, you too have a fire-breathing supercomputer just ready to go to work for you. You’ve already paid for it.
On the desktop, you can plug in a couple of NVIDIA GTX 295’s and get, what? over 500 stream processors clocked at desktop speeds, far beyond the power of some goofy old laptop that has to run a long time on battery. You can get over a teraflop of computational performance, so fast that it would have cost over a million dollars just a few years ago when ESRI started building ArcGIS Server. And you can buy all that today for well under $1000.
All I am saying is that it is criminal neglect not to build GIS software at the desktop, because even the dumbest desktop when you add it up as distributed power across just a handful of users delivers far more GIS power than whatever overpriced server ESRI wants you to run ArcGIS Server on. And it’s not a small win either, it’s a huge win, often with hundreds or thousands of time more computational power.
I’m not saying there isn’t any role for server side software. Of course there is, and on the job I work for a company that builds server-side software as well as desktop software. But for most people who do GIS day in and day out for eight hours a day the best architecture is to leverage servers to centralize DBMS and data warehouses while distributing GIS functionality out to what have become massively powerful desktops. You’re paying for those desktops anyway, so may as well use them as full power computational engines for all you can wring out of them.
Even if you don’t do any of the gee-whiz CUDA supercomputing stuff, it’s now becoming the case that with Intel Core i7 you are effectively getting eight cores in action, so there’s a lot that multicore desktop GIS software can do for you. If you think i7 is too new to consider, the inexorable power of the mainstream means that within about three months it’s going to be what you see offered at Costco for your college kids to buy next semester, and within six months it’s going to be the discount special for your high school kids coming back from summer vacation. Today, it’s already cheap and standard for your professional use.
I grant that only recently has Core i7 arrived, and that maybe it’s fair to say that going from quad core to eight core is new. Very little software takes advantage of that, not even some of the most modern GIS packages do quite yet as they could, and it is phenomenally difficult to code heterogeneous parallel architectures. So what? Isn’t that what your GIS vendor is supposed to be doing for you, busting a gut to make it possible for you to take advantage of the latest and greatest hardware at the lowest cost so your work can go easier and smoother? The point of them working hard on your behalf is so you can take advantage of the best technology for your interests today, not five years after it has become obsolete.
Let me end on a cheerful note, to reply as to how to end the strangle hold. I’ve yet to meet anyone quite as effective as getting what they want as an experienced Federal procurement officer. There is always a way, and usually that way is made easier when what you want is something that is faster and better. I mean, people may look at you like a nut case if you start talking about getting a better price, but if you just say you want this thing because it is the fastest and the most powerful, well, anyone can understand that. I hear that’s how a lot of the new, modern GIS pacakges are finding their way into various agencies. Users just demand there’s no way they will settle for 32-bit desktop software on their 64-bit desktop systems. They want the best, and they’re not going to spook anyone by mentioning that it is lower cost as well.
Er, I’m no expert, but isn’t ESRI’s server-side shift partly a response to the massive growth of web-mapping client tools in recent years, and the desire of many customers to make at least some of their GIS products (in the looser sense) available via the web? They seem to be trying to compete simultaneously if somewhat awkwardly in very different markets – high-end professional GIS analysis using thick client software (where they’ve always been strong) and the newer market for consumer/prosumer browser-based web-mapping applications. Lots of people seem to like the new ESRI web stuff with ArcServer, but I still don’t understand why ESRI can’t make its professional client software talk to a proper database without ArcServer. They still seem stuck in the proprietary silo/stack mentality to me. Another topic altogether, I guess.
As much as I find your view on the bailout mistaken, I have to give you credit. You’re right on about many technical matters here. But the beauty of ESRI is that we reinvest heavily in development, so you never know what (desktop) innovations are just around the corner.
I sure hope that is true. But, Dimitri is sort of correct in that ESRI hasn’t done this yet. I’m not sure why? Can Jack get 10 guys, lock them in a room and say: “don’t come out until you have this thing working in 64-bit, multi-processing, etc.”.
I mean couldn’t he just start out with the raster stuff. My guess is with 10 computer scientists working on this for an entire year, they should be able to do it.
Remember, alot of ArcGIS is a toolbox, with independent tools (overlay, buffer, etc.). How long would it take to convert the raster overlay to 64-bit. Lets assume that each programmer can take 1 function and convert it in a week to 64-bit and multi-processor. That means with 10 programmers they can have 10 functions done in a week. In 50 weeks, thats 500 ArcGIS commands that are multi-core and 64-bit.
Jack could easily spring for the salary of 10 programmers to do a back-room rewrite, right?
ESRI is already adding 64-bit and multiprocessor to their server and desktop.
Thanks for the info James. Better late to the party than never
So Dimitri, does this mean you won’t be able to call ESRI legacy software? They’ll be 64-bit, multi-parallel to boot. And, they have the programming staff to churn this out quickly.
Would this level the playing field?
I’m going to un-indent this a bit by replying to a prior Harpo post…
Harpo writes:
“My guess is that ESRI has 3 times the programming staff, and in one year, they can do more than these small companies.”
That’s an understandable guess that many people make, but it is way off the mark. Understanding why it is off the mark can help explain outcomes that otherwise seem baffling.
The key factor is not how many people you have overall, but how many people you have actually doing core development. An important corollary of that is not just the absolute number doing development but what percentage they represent of the overall headcount. Almost everyone immediately grasps the importance of raw numbers doing core development, but few who have not run technology companies recognize that the percentage is critical as well.
A brontosaurus was one of the largest land animals ever, far larger than humans, but humans have a much larger brain than the walnut-sized brain of a brontosaurus because humans devote a much higher percentage of their body mass to brain and not bulk. The greater brain size of humans lets us do things like appreciate Shakespeare and write GIS applications, while a brontosaurus no doubt pretty much ran on wired instinct when approaching the challenge of deciding which end of a tree was the leafy part.
But the higher percentage of brain in humans also drives a very different survival strategy than the brontosaurus. It fundamentally changes our behavior in endless ways large and small, from autonomic physiological tactics that keep blood flowing to the brain before other organs in times of stress, to large scale strategies such as building civilizations and science.
The surprising thing to many people who have not been inside the development world of a large company is how very small the core engineering teams are even for very large products. It’s often the case that something like a new spatial DBMS is created by a handful of guys, literally only five or ten or twenty, even within very large companies that have tens of thousands of employees.
If your company invests most of its headcount into engineering, a surprisingly small company can end up having a significantly larger engineering team working on the core product than even a very large company that invests most of its headcount into activities other than engineering. Given the habit of large companies to overweight internal political matters, it’s also often the case that what resources are invested into engineering get invested less efficiently than they do in smaller companies where top management has fewer layers separating them from the actual fighters. Read the recent Wall Street Journal coverage of Microsoft’s missed opportunities in paid search for a classic example of that.
I can’t speak for the other modern GIS companies, but the modern GIS company with which I am most familiar puts 95% of headcount into engineering on the core product, with all other activities run on just 5% of headcount. That’s why they don’t do any sales or marketing or advertising or send idiots out to help make designs by committee or respond to Federal RFP’s. It’s all engineering, darned near as close to 100% as can be. I have reason to believe that the other new companies are also more focussed on engineering as well.
ESRI, in contrast, probably has the inverse proportion with only about 5% of its headcount invested into guys who actually develop the core product and 95% of it spent on all the other activities on which ESRI spends its time. Like seeking corporate welfare from Uncle Sam.
That low ratio of actual engineers makes it very difficult to attract and retain the very best engineers (hotshots don’t like being a part of an organization where they don’t count for much) or to manage them well (too much mindshare of senior executive staff has no choice but to be applied elsewhere). It’s the classic brontosaurus model: tons of bulk wrapped around a walnut-sized brain.
I have some hard data on that as well, because due to an accident of timing I happened to be present at the time my company realized they were dealing with a walnut brain as an adversary.
A few years ago Microsoft was trying to convince engineers to keep alive a technology that was obviously a failure (Visual Basic for Applications) so they set up some seminars for engineering representatives from hotshot companies they wanted to talk into using VBA. About 300,000 applications used ActiveX or .NET scripting but fewer than 150 at that time were using VBA, despite years of effort by Microsoft to promote it. Microsoft was making one last effort before pulling the plug.
Some of our engineers were asked by their friends at Microsoft to attend to help bulk out the crowd and they felt compelled to attend. I ended up going since we wanted our delegation to be as big as possible to show support, even though no one from engineering wanted to go.
Much to our surprise, Microsoft’s way of pitching VBA was to trot out engineering folks from whatever companies they could muster who had actually used VBA to praise the value of it, including ESRI’s lead engineering manager in the team that did Arc 8.x, at that time ESRI’s first true code for Windows.
We were surprised to see the presentation talked very openly about ESRI’s challenges in adopting Microsoft ways, learning MFC and Studio and all that stuff and ultimately going with VBA and also disclosing the actual numbers of people involved in this at ESRI. We were very surprised to learn that ESRI had fewer actual developers working on their core product than we did, although on reflection that makes sense given that they have a whole lot of legacy nonsense to deal with that sucks resources out of core development for new and improved products.
The VBA presentation also fell flat, since of course you don’t impress hotshot engineers from companies that have been doing Microsoft stuff as an intrinsic part of their lives for many years with a presentation by someone who is still trying to learn Studio. But that did make an impression on us as well, because it was clear that the reason ESRI’s products had so many dumb, newbie mistakes in them is that these were really a first attempt for them using what was for them a new set of technology tools. Not bad at all for a first attempt, probably no worse than our first attempts were, but given that they were competing with people who were deeply into fifth and sixth generation use of that technology in comparison they were indeed newbie mistakes.
Anyway, our takeaway from that was that ESRI was very vulnerable to an attack based upon better engineering. The outcome of that day, along with many other factors of course, was to convince our company to make the investment required to grow our GIS line from a relatively simple vector-only GIS into something that would combine vectors, raster images, terrain elevation surfaces and do so at the full enterprise level with integrated IMS and spatial DBMS.
One of those factors also was not just that we knew we could out-muscle them in raw engineering throughput, it was because the history of ESRI’s struggles with new technology made it clear that expert usage of technology did not have the executive mindshare that was required to fend off a competitive threat from people who knew how to use the technology weapon in combination with other economic and business factors (can’t just win with a better mousetrap, you know).
You can see how that’s played out, because the kind of company that puts most of its effort into lobbying isn’t the kind of company that will put sufficient effort into, say, going 64-bit.
The bottom line is that thinking a larger company can just throw people at a problem to make up for lost time isn’t how development works at large companies. They often don’t have as many people as smaller companies which do focus on development, those people often are not as well managed because managers must share mindshare with too many other functions, and then finally you have the problem that it is not so easy to elevate the game of weaker teams that result when you’ve spent many years making that a lower priority. That’s why companies in this fix often deal with it by trying to buy a more effective organization instead of reforming their own. Microsoft, for example, is pretty much writing off its internal search team by saying they need to buy an external search company like Yahoo. [Although I grant that example is not just about engineering.]
Wow, lively debate. Begs the question: If Google had written this proposal, would we all be bi$%&#$ about it, or would we be joining hands and singing its praises?
Gotta tell you, I don’t care who gets the money, either way we’re (US taxpayers) are screwed in 10-20 years when the bill comes due… and at least some of the bailout money will be helping my 401k out with the part investing in IT firms (Pitney, Microsoft, etc).
Plus, if we all read the whole proposal, you’ll see things like imagery components (helping Leica & others who are near collapse), so in the end it just might help all us GIS/Imagery folks in some way, who knows.
My 2 cents. -GT
I was half joking in my last post, but apparently Google and Integraph added a similar proposal:
http://industry.slashgeo.org/article.pl?sid=09/01/29/1347200&from=rss
Now, are we all going to bash THIS as well??? Or since it’s the almighty Google, will we all do a 180 and sing “Kumbaya”?
Pork is pork, no matter who the author is! -Geotom
May be Google can foresee the time when it might need to ask for a bailout too.
Someone remind me what of tangible intrinsic value Google actually makes?
Since the start of 2008 the value of Google stocks has almost halved and they are very probably still greatly overvalued. Its continued high value is dependant on market confidence that it can continue to make piles of money during what is shaping up to be the worst global economic recession in history, when businesses are either going bust and those that havent are thinking carefully about cutting costs in all areas, including targeted online advertising. Lets not forget Google’s service products are simpley vehicles for commercial advertising and so Google is primarily an advertising services company. Like all service companies it is highly dependant on a bouyant economy to provide the demand for its services. If this demand shrinks significantly Google may be in trouble. In many ways it is a relic of the dot.com bubble that has yet to burst.
On a general note, the proposals for investment in US national geospatial infrastructure, if implemented, may indeed stimulate the US economy and may be the world economy too, but only a little. This recession has been caused by greed and risk taking in the financial industry (securitisation of junk loans mainly) and has since spread because of the unwillingness of banks to now extend credit. The proposal is rather like replacing a worn tyre on a car when the transmission has clearly failed. It does not fix the reason why the car will not run.
Moreover, because of the hi-tech labour unintensive nature of this sector of industry, the proposals are unlikely to generate many jobs or guarantee job security for many. Indeed many of any jobs created might be abroad. Instead the proposals look rather like a cynical opportune means of channelling large amounts of public money into a few select private pockets.
Have to agree with you on that, Shrek.
The Google folks aren’t stupid. They understand that when you take money from the poor and put it into your own pocket it’s more important than ever to cloak yourself with “Do no evil” nonsense slogans.
Google makes money at one thing: search ads. That’s it. They lose billions at everything else they do. They know that better than anyone and they don’t mind taking advantage of any opportunity to reduce their costs by looting the treasury. Everyone’s doing it now, don’t you know, and Google has already had a taste.
A good example of that is Google Earth, which was based in large part on the notion of privatizing the many billions of dollars’ worth of satellite imagery collected by Uncle Sam using public funds. [Yes, you bet Google showed initiative with GE and good for them for doing so. But the actual technical content of GE is about a week's worth of serious engineering for a small team: the prime value of the thing is the billions of dollars in satellite imagery it displays.]
I mean, no one in those space / remote sensing / big science Federal bureaucracies actually gives a hoot about anything but perpetuating their fiefdoms, so tons of data accumulate that the public won’t be allowed to have, until someone big enough like Google comes along to corrupt the right people. Oh, excuse me there, I forgot in Washington it’s not called “corruption” it’s called “public / private partnerships” or “cooperation with stakeholders” or some such facile idiocy.
So all that fabulous imagery that your tax money paid to accumulate and which was kept under lock and key to prevent you from using it got… privatized. All of us with extensive Washington experience know how that works. The feeding trough is for “stakeholders,” it’s sure as heck not for the smelly tourists, as our distinguished Senator Reid likes to refer to us taxpayers.
For the tiniest fraction of the money already budgeted all the agencies with remote sensing data could put it online for free download. They don’t do that because they really don’t give a flying hoot about giving the public access to what the public paid for. They’d rather pull down a few billion to study the problem and then spend a few billion more to make sure no one gets access to that data except the lobbyists and their “partners” that have corrupted them. Business as usual.
All that’s a shame because the wide availability of geospatial data does indeed help power the economy upwards. If it weren’t for the massive amounts of free data USGS put online for free download there would have been no way for the current crop of modern GIS companies to have gotten started, and thus there would have been no way you could get the most sophisticated and most modern GIS in the world for under $500. Widespread availability of free data has made it possible for millions of people to be using inexpensive GIS for all sorts of business and social gains. Considering how that has been accomplished with a fraction of public Federal data being made available to the public, it’s easy to see how uncorking the rest could accomplish even more. But you don’t need multibillion dollar payola for Google to accomplish any of that.
Please, let’s all be worldly enough to understand that a Federal plan to shovel billions of dollars into subsidies for Google has nothing to do with wonderful civic infrastructure growth or public access to wonderful new public data or any other such fast talk aimed at rubes. It’s just Google realizing that when the greatest looting of any public treasury in the history of the world is going on, well, they may as well help themselves to some of the pork. I guess from a worldly political point of view that is fair enough as political payback for Eric Schmidt’s support of the winning candidates – that’s how politics works, don’t you know. Elections count and to the winners go the spoils.
[Those who think Google's intent is public access to public GIS data should ask someone with legal savvy to explain to them the meaning of Google's terms of service.]
I have to say, though, that I feel so exhausted by the total corruption of trillions of dollars flying out the door that a few billion for Google doesn’t seem all that awful. I mean, they do have a cute logo and the search engine is pretty good and they have served an important public service in teaching Microsoft that it is not realistic to base a business on annoying your visitors (that whole obnoxious display ad thing as compared to discrete AdWords ads).
And, despite Google’s cat and mouse terms-and-conditions game with Google Earth users and their occasional jihads against nice people like the NASA WorldWind team, well, it’s not like it takes more than a middle-school level of programming talent to use whatever free data you want from their web services. So maybe they’re not so bad and perhaps if you are going to hand out billions for porn and people who buy houses with no intention to pay their mortgage, may as well find a few billion for poor Google as well.
The Keynote at MAPPS last week was given by Mark Jones with Google. He had great things to say about our favorite low cost GIS software company …and nothing to say about the 800 pound gorilla in the room. The mantra from the ESRI godhead that data should be free is eventually going to be true. What he (Jack) didn’t foresee is that software/bloatware is going to follow data down the same path. It’s what you do with data that has value.
Dmitri,
One point: “Google showed initiative with GE and good for them for doing so”
Google didn’t show any initiative, they just rewrapped an existing product called Keyhole Viewer, it sprang out of an In-Q-Tel initiative from years ago (http://www.iqt.org/).
Cheers, GT
All absolutely true, I’m sure, and thanks to our own Ordnance Survey here in the UK we know all too well how public data suddenly morphs into private data: we feel your pain!
But the thing about Google is that, while lots of important players could have made all this geo-data easily available to Joe Public (and could have done it better or more cheaply or whatever), the truth is that all those GIS experts and specialised GIS software companies and politicians and clever tech start-ups and corporate IT leviathans never did it. Google did.
Now you guys all know the value of more sophisticated GIS tools and underlying data, but to most politicians and the ordinary public, it is Google that is the face of internet mapping technology, and it is Google that one way or another has helped to generate a public interest in SDI and (perhaps) the political will to invest in SDI. Because without all the PR and BS and simple fun around those Google web-mapping toys that Dmitri says could have been hacked together in a few weeks (which raises the question: why didn’t the Manifold boffins do it in their lunch hour?), do you really think the public would want to spend a billion dollars or more on a spatial data infrastructure in the current climate?
I’ve no special axe to grind on behalf of Google, which is of course acting in its own interests like any other company, but all this bitching by GIS industry insiders seems to neglect the fact that Google’s approach has helped to inject new life into that industry by giving Joe Public at least a vague understanding of the potential of the SDI that he is likely to end up paying for. After all, how much of the current GIS industry in the US would still exist if you took away all the public money, from county planning, national infrastructure, natural resources or defence? You need taxpayers and politicians who can appreciate the importance of spatial data infrastructure. Google’s contribution has at least made that task of communication a little easier.
Anyway, good luck grabbing your own shares of the geo-pork, guys!
Geotom writes…
“Google didn’t show any initiative, they just rewrapped an existing product called Keyhole Viewer, “
That’s what I meant by initiative, that they did that and connected it to a large infrastructure of servers loaded with data and brought it out for the world to use. It wasn’t any real software development but it was a clever and insightful thing to do and getting that data did indeed take initiative. I commend Google for that. A good job well done.
ChrisW writes,
“Now you guys all know the value of more sophisticated GIS tools and underlying data, but to most politicians and the ordinary public, it is Google that is the face of internet mapping technology, and it is Google that one way or another has helped to generate a public interest in SDI and (perhaps) the political will to invest in SDI. Because without all the PR and BS and simple fun around those Google web-mapping toys that Dmitri says could have been hacked together in a few weeks (which raises the question: why didn’t the Manifold boffins do it in their lunch hour?), do you really think the public would want to spend a billion dollars or more on a spatial data infrastructure in the current climate?”
Chris, I agree with you on quite a few points but not on the above. I think you may have misunderstood my comment as well and that we could be in violent agreement as regards data/software.
I think we both agree that the value in Google Earth is the imagery. It’s not the software for viewing it. In point of fact, there have been many individuals that have cobbled up viewers for such things in a weekend’s work.
To take a specific example from the company you named, their toolbar plugin for Internet Explorer, while not quite written over a lunch hour I think took not more than a day or two. I’ve heard of dozens of different people who have written image server modules for that toolbar or for the GIS system which inspired it and I don’t think any one of them took more than a day or two. See http://www.manifold.net/toolbar
The NASA WorldWind experience is also instructive: it’s not hundreds of guys working for years that’s required to do this. So, yes, I think my comment that a reasonably cool viewer is about a week’s worth of serious engineering for a small team (say, fewer than a dozen experienced people) is scaling it about right. Whether we’re talking about a week, or a month or even two months, that’s insignificant cost for just about anyone in this business, let alone for Google.
In contrast, the data displayed by Google Earth comes from hundreds of billions of dollars of mainly Federally-funded infrastructure and activities on a massive scale over decades. Add in the indirect costs of launch facilities and the like and it is truly an epic expenditure.
Google is now talking about launching their own satellites, but when Earth started that data was already “in the can.” It just wasn’t available to you or me or to any company not willing to venture the fairly massive interaction with the Federal bureaucracy that’s required.
As to making that data available in a bona fide “public” way, you should take a close read of the Google terms and conditions of service. You’re allowed to view that public imagery served by Google only if you use Google’s browser or Google’s SDK (as if a proprietary SDK is somehow required to parse an HTML string… that bit is air cover for the technologically illiterate) and even then your ability to use it is grossly restricted. No way, no how, is this provision of public data for public use. It is a private advertising vehicle kept on a short leash.
I’m not criticizing Google for that, I’m disagreeing with the implication that Google is providing public data for public use. They’ve privatized public data and are using it for private gain. That’s OK and perfectly allowed and good for them for showing the initiative to realize they could do that. It even has positive public benefits despite being a very private advertising mechanism. Seriously, good for Google for doing this.
But it’s not at all a remedy for the Federal agencies which kept that data hostage. It is the Federal agencies which are being crooked here – I don’t really blame Google for dipping a hand into the public trough. At least in that way some of the imagery sees the light of day.
But it’s that hostage effect that should be kept in mind if anyone is tempted to shovel billions of dollars more out of the treasury. Give that to agencies that kept the imagery already created under lock and key and all you will assure is that whatever new data (if any, a long stretch at that…) is created will also be held hostage until it finds its way through a crooked seam into a privatization scheme.
As far as anything Google does impacting the public’s willingness to fund SDI, well, the public “votes” for that every time they click on a search ad published by Google. If this makes sense for anyone or anything it will be profitable for Google to fly their own satellites and generate their own imagery.
But the public’s interest in consuming text advertising on Google sites doesn’t tell you anything about their willingness to finance any real spatial data infrastructure, because Google’s privatization of consumer grade imagery doesn’t really serve the constituency of people who do GIS eight hours a day.
If anyone is really serious about wanting to use Google Earth-like applications for improving the efficiency of our economy by leveraging the government’s investment in spatial data infrastructure, well, you don’t have to structure a multi-billion dollar subsidy for Google’s private advertising efforts. All you need do is a) compel the agencies to provide public access to their existing imagery and other public data – that’s millions not billions, a drop in the bucket of their existing funding, b) make it unlawful for anyone who redistributes in any way public imagery to claim ownership of that imagery or to restrict subsequent use of that imagery. You know, the old “copyleft” thing so beloved of open source types.
There’s plenty of data already. The problem is that it is not be made available to the public.
Note that none of the above prevents Google or Microsoft or anyone else who thinks that providing access to public imagery is a good advertising vehicle from doing so, nor does it prevent them from flying their own satellites if they think public imagery isn’t good enough.
I’d also tell the Feds to cut Google and other would-be remote sensing operators some slack – if Google or anyone else wants to fly a satellite with six centimeter resolution and launch it on a recycled Russian booster off the coast of Africa to save cash and get it into orbit quicker, don’t stand in the way of technological progress and willingness to take investment risks.
One last observation for anyone who thinks this has anything to do with economic stimulus: none of this will have any work product available for years. All it will do this year is allow an additional wave of bureaucrats to be hired in Washington, and there’s no unemployment problem at the Federal agencies. Any spatial data infrastructure improvements that come out of this which will have any benefit to the public or have any impact on real job creation aren’t going to see the light of day until 2011. In other words, a massive subsidy in this thing is going to have zero effect on job creation or economic stimulus in 2009. Mark my words, comrades, this is no way to fend off another great depression from kicking in during the next six months. All it will do is suck what money there is out of productive purposes where it is needed this year and catapult it in flames into some distant year where it will land as ash, accomplishing nothing but crushing people with higher taxes, further debasing the currency and adding the misery of inflation to a more extended depression.
Hi Dmitri – thanks for your comprehensive reply. You’re right about all the data and real-world GIS issues, as opposed to consumer web-mapping, of course, although the privatisation of public assets is only possible if government and to some extent the public allow it to happen. Blaming companies for cashing in is like blaming sharks for eating fish: it’s just what they do. As for Google, I don’t think they’re any more or less “evil” than other companies, although they did kill Bambi:
http://www.guardian.co.uk/media/pda/2009/jan/30/google-digitalmedia
Cheers, Chris
I agree with some comments posted on the All Points Blog http://apb.directionsmag.com/archives/5255-Update-4-ESRI-Invites-Support-for-GIS-for-the-Nation-as-Part-of-Stimulus-Bill.html
I think the proposal dramatically understates the costs that will be involved in areas like creating a nationwide parcel database. The scope of this project is sufficiently broad and vague that it really needs to go through a rigorous study phase before any money is spent on implementation. If the government wants to build a targeted GIS application for internal use for monitoring stimulus projects, so be it, but rushing out to spend 1B dollars (conservatively) on a IT project of this magnitude is going to lead to unsatisfactory results.
I heard a story from an insider… years ago, most likely true… Apparently, Jack had just wow-ed some big client, with big cash, and big hopes, and sold them on the cool aid soaked T shirts and the whole ssshbang. After the spiel, he gets in the car with the sales guy travelling with him. Sales guy says something to the extent of “Wow, that was incredible, I’ve never seen anything like that.” Jack looks over at him and says “It’s all about the money…drive.”
Let’s face it, when Jack’s dead, we’ll get a more open industry. Until that time, it’s pure showmanship. Jacks’s an artiste. I bet he’s gunning for that Fed cash just because he thinks he deserves it, not because he needs it right now. I did think that piece was beneath him, though. Maybe he’s hitting senility? I truly don’t think he gives a crap about parallelism, fat client, thin client, server based whatever… He’s running a cult out of Redlands that aspires to the bottomline at the expense of significant progress as found in other industries. Why? Cause he can…for now. I’m really not upset about it like I used to be. I think I will go check out some Manifold now… BTW, if you’re going to give any money to the USGS, you might as well just burn it. They’re lost.
You guys are great! I love the concern shown for ESRI. I’m sure Jack and company are sleeping better knowing that all of you are coming up with a way to save them from their impending disaster. If they aren’t sending chocolates right now, they should.
Admin Note
At Wordpress 2.7, there was an ability to add nested (threaded) comments. This seemed to work well until this thread and everyone replying to everyone else. I changed the way this works, but of course this thread is now more confusing than ever before (or maybe it makes more sense).
Back to work all you Nostradamuses!
I had one thought in the back of my mind while reading Dmitri’s ideas about modern vs legacy GIS. Where do you guys work? I still have to deal with Pentium IIIs, occasionally a Pentium II. I don’t know of a single person in my section of the organization (which includes computer services) with more than 2GB RAM on their desktop, and the only computers with add-on graphics cards (almost all standard PCI bus) are the ones that need two monitors. My best in the section computer is a 3Ghz Pentium 4 with 1GB of RAM and a AGP FireGL V3100, which will be up for replacement in two years.
64-bit desktops with 8 GB of RAM, multiple graphics cards? Well, in 2 years I might get to spend $800 to replace my desktop… It’s that crazy world of government budgeting. I can “spend” $1500 for the ESRI UC because it was in the budget cycle two years ago (even though I get a free registration), but I can’t “spend” $70 to go to the state GIS conference because it wasn’t. There is no chance I am going to get the $469 to upgrade my desktop memory (yes, I have to get the vendor’s upgrade memory) to 4GB before 2011.
And the worst part? We’re a high population county; one of the cores of ESRI’s customer profile after federal government. I am not familiar with the federal situation, but state desktop computing resources look to be in worse shape than us. The money goes into software and servers (did I mention our core servers run unix?), with a steady shift towards citrix and vm environments for desktop users.
What Atlas cannibalized or what it cannibalized and embedded in the core ESRI offering?
Supercomputer use 64 bit since 1960 (wikipedia) so osftw
You’re all wrong. Jack couldn’t give two peach pits in a pot of piss about money. For Jack, it’s all about control.
To paraphrase his associate Mr. Ballmer: Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share, Market Share
Money’s a (very) distant second to the man with a vision.
“You’re all wrong. Jack couldn’t give two peach pits in a pot of piss about money. For Jack, it’s all about control.”
If he thinks this, why are ESRI products so stratospherically priced, yet so technologically obsolete?
Well define your definition of “legacy software”, then decide what ESRI and Manifold are.
http://www.spatiallyadjusted.com/2008/01/31/legacy-gis/
I find the whole legacy vs modern software issue to be a bit of a red herring. The real question is whether or not the code/program/platform meets your needs. For example we would all quite likely classify FORTRAN as a legacy language and yet old FORTRAN code is what is used in much of scientific computing (e.g. weather simulation, nuclear simulation). Better yet you can enhance your “legacy” FORTRAN code with MPI or OpenMP and run it on massive computing clusters.
A more appropriate and accurate term to describe ESRI’s current range of products is “obsolete”.
They are obsolete because ESRI products can not utilise the latest software and hardware technology, as Dimitri so eloquently explained.
Shrek,
read the thread. James just said that ESRI is going to support multi-parallel and 64-bit. Is that obsolete?
Harpo,
I referred to “current range” not vapourware. However, even if ESRI products will soon support 64-bit OSs and parallelistion, modern GIS software has a (let’s be generous to ESRI) five year lead.
Since the term obsolete describes a state an item has relative to a more modern item, the term obsolete will still apply to ESRI products even if these hypothetical future products support the latest version of 64-bit Windows and begin to provide some parallel processing capabilities tomorrow. In the mean time, the technological lead that the most modern GIS products now have over ESRI ones, will be maintained and will very likely grow at an accelerating rate. For ESRI to catch up it has to increase the rate of the technological advancement of its products at a rate greater than that of currently more technologically modern competitor products. Or, modern competitor products have to decrease the rate of their technological advancement. Or, both have to occur together. It is possible for ESRI to catch up and it may still, but it looks increasingly doubtful. I am sceptical that it will do by any way.
The damage done to the reputation of ESRI, by their failure to adopt early, the-then latest technology, has been great and is growing daily.
The 64-bit OS, multicore CPU and GPGPU CUDA/Open-CL parallisation train departed the station many years ago. ESRI failed to get aboard and is now left staring down the track as the last carriage vanishes over the horizon.
Basically, they blew it.
James wrote
“ESRI is already adding 64-bit and multiprocessor to their server and desktop.”
I don’t doubt James’ word at all, but the above could mean a lot of different things.
For years ESRI said their products ran in 64-bit Windows in what to my taste (I realize tastes differ on this) was a slippery and deceptive way.
I talk to ESRI users all the time who were under the false impression that the ability to run a 32-bit desktop ESRI product within 64-bit Windows was indeed “running 64-bit.” That’s not true.
Sure, 64-bit Windows can run 32-bit applications, like 32-bit ESRI desktop software, but it does so by in effect forking a 32-bit Windows environment. So when you run 32-bit software in 64-bit Windows you don’t get the benefit of 64-bit operation. It’s plain, old 32-bit Windows with all the limitations, such as grossly limited process space, thereof.
If your GIS functionality depends upon the interplay of many products, it is enough just to make some of those products here and there 64-bit. You have to make the entire stack 64-bit if you really want to utilize the benefits of 64-bit Windows. Given that ESRI is struggling to make this happen for a limited set of products and taking years for that, I’ll believe full support for 64-bit operation across their entire product line when I see it.
Regarding the multiprocessor thing: that is a far more technically difficult thing than going 64-bit, and there are ways of saying one supports multiprocessor operation when what is being done is a very limited thing.
It’s easy to do things like write multithreading for servers. No big deal: any new session that gets launched runs another thread on another core.
You can also do that with data access to DBMS fairly easily, and with a bit more effort to file access within interactive settings like the display of tiles from image library files.
But it is extremely difficult, something that has defied the best efforts of the world’s best software companies, to implement true parallelism where a computational task is automatically decomposed into multiple threads for parallel execution on multiple processors and then the results reassembled.
It’s not just a matter of coding, either, as difficult as that is. One often has to invent genuinely new algorithms, that is, bona fide new mathematics, to parallelize a task in a way that you know the result is accurate.
But achieving that decomposition of intrinsically single threaded things into distributed, parallel things is the whole point of getting past Moore’s Law using parallelism: it’s not in assigning out what are in effect single-threaded tasks to multiple threads, instead it is being able to take something which is intrinsically a single thread and decomposing it out for distributed computng and subsequent reassembly.
Let me make clear the difference by making an analogy to a retail bank. Suppose you have a bank with a row of teller windows and a line of customers, each with some business to do with the bank.
If each customer has something simple to do, like make a deposit or withdrawal or ask about a loan, the simple multithreaded case is like sending each customer to a teller. If you have ten tellers and ten customers each customer goes to a teller and does his or her business and the line gets worked down.
That’s like the simple multithreaded case, like visitors to a server-side web site. Each user wants one simple thing, to see the site, so you can have a bunch of processes more or less doing the same thing but on different cores.
The parallel case is much different. Suppose you have just one customer in line and that one customer has a big job, like reconciling a year’s worth of accounts. Follow the simple multithreaded model and that customer walks up to one teller and works with that one teller however long it takes, say an hour, while the other tellers are not used.
The more complex case of parallelism would be like if the customer could step into line and automatically his or her job was sliced up into ten different parts, perhaps each teller somehow taking one tenth of the accounts, and then the results reassembled. In that case, instead of just one teller working on the job and nine others sitting around doing nothing, all ten tellers go to work and the job gets done in six minutes instead of in an hour.
This is like someone who wants to reproject a vector drawing, or perhaps do a sophisticated spatial overlay: it’s one person at his desk who wants the job chopped up for all four cores or eight cores or whatever to go fast.
It is extremely difficult to implement such parallelism in a GIS. You end up having to re-write tens of thousands of routines so that each is parallelized.
That is extremely difficult to do and, to date, no one has ever successfully done that in a full scale application the size of a big time GIS. Further, it takes years of intermediate steps to learn how to do that well. The first efforts of newbies at parallelism will often run slower, not faster.
My roots in this go back to the 1990’s when at Intel I helped put the team together that did the massively parallel mathematics libraries for Intel’s Paragon supercomputer that came out of their joint project with DoD (Touchstone Delta). Since then I’ve seen a lot of groups take on parallel code.
In recent years, I’ve reviewed the work, read the papers and in many cases talked with the developers of over one hundred teams that have implemented massively parallel algorithms for the likes of NVIDIA GPU technology.
These are really, really smart guys from the most advanced parallel coding teams in private industry, academia and government and three things stand out from all that.
First, it takes about a year for even a very talented team to get past the newbie stage.
Second, even experienced teams take years to implement more than a handful of functions. This is very difficult work, years in the making.
Third, while one occasionally encounters a GIS company or remote sensing company, ESRI has been totally absent from all that.
My conclusion from all the above is that when James reports plans for multiprocessor usage at ESRI he is probably talking the simple multithreaded case and not parallized processing.
Writing 64-bit code is very easy, journeyman development that is far, far easier than writing parallelized algorithms and code. Given that ESRI has made but glacial progress at the easy things I don’t believe they will make any faster progress in the truly difficult engineering tasks.
A typographical error, missing a NOT which should have been there. Apologies to all. The corrected text:
“If your GIS functionality depends upon the interplay of many products, it is NOT enough just to make some of those products here and there 64-bit. You have to make the entire stack 64-bit if you really want to utilize the benefits of 64-bit Windows.”
“For example we would all quite likely classify FORTRAN as a legacy language and yet old FORTRAN code is what is used in much of scientific computing”
Doug, I hear you and I appreciate the sentiment. But you’ve picked a good example of why the word “legacy” is sometimes a pejorative.
Sure, there is lots of legacy FORTRAN code out there and just as surely that’s why people grind their teeth and curse the high costs and inefficiencies of dealing with legacy stuff.
Ever try to hire a FORTRAN programmer? It’s probably about a thousand to one odds against you compared to VB or C++ or C# or Java given that programmers these days are almost universally trained and experienced in other langauges. If you can find a guy who has FORTRAN chops it’s likely that he or she isn’t all that up to date regarding modern development environments or languages.
I’m not bad mouthing FORTRAN, I’m just being respectful of the “march of time” issues involved that tend to push developers into one or another generational cohort.
Regarding progress, sure people have invested efforts to use FORTRAN with modern hardware. But that’s just a fraction of what more modern languages enjoy. It’s not like every hotrod young entrepreneur who’s on fire with a new idea for a cool new compiler is going to choose FORTRAN to do first.
I agree that if something works well for you and it is already paid for, well, why change. But that only goes so far because you might be missing out on lots of benefits from newer technology.
modern GIS software has a (let’s be generous to ESRI) five year lead.
no, for a couple of reasons:
Manifold hasn’t had multi-processor or 64-bit for 5 years. Maybe two.
Also, its not a full two years. Its how ever much programming they can do in 2 years. My guess is that ESRI has 3 times the programming staff, and in one year, they can do more than these small companies.
Harpo wrote:
“no, for a couple of reasons:
Harpo,
I was careful to be vendor neutral, but since you mention Manifold[.net], you should know that the Company and the Manifold System user community are very well informed about technological trends in IT hardware and software. The Company knew (as indeed anyone with any common knew) well before 64-bit CPUs/ and 64-bit OSs were released that these were the future. On this basis Manifold.net evidently had the foresight to take the strategic decision to develop its products for the coming technologies, not those in use then or yesteryear’s. Therefore, its 64-bit software products were ready for these hardware and software inovations very soon after they became generally available. So yes, I stand by my ascertion that Manifold.net has a long lead over ESRI in this field, just as it now does in developing its products to take available of multicore CPUs and the massive potential of GPGPU technology.
FYI the first general purpose mainstream X86-X64 CPUs were manufactured by AMD and released, as the first in its on-going product line of Opteron CPUs, on April 22, 2003. Shortly thereafter, Microsoft who had been working closely with AMD to develop an 64-bit Windows OS to run on Opterons, had a public beta 64-bit version of Windows XP available. Since we are now at the start of 2009, Manifold System has more than a 5 year lead (Jesus, more like six years!) on still-32-bit ESRI products. In contrast, ESRI, it would appear, is still stuck in the late 20th century, technologically.
Some of the best raster analysis I’ve had done for me in the past year has been with fortran. Legacy is only what you make it.
Dmitri:
“It’s often the case that something like a new spatial DBMS is created by a handful of guys, literally only five or ten or twenty…”
Totally off topic, but that just happens to be the ideal size range for a group where you can hope to get agreement over decisions:
http://www.newscientist.com/article/mg20126901.300-explaining-the-curse-of-work.html
This research found that groups tend to splinter once they go above 20 members, and for some reason, a group of 8 members is particularly prone to deadlock. Guess all those small Agile teams might be onto something, eh? Hope the US government bears this in mind before they plough $1bn into one mega-project…
That’s why companies in this fix often deal with it by trying to buy a more effective organization instead of reforming their own
so, when is ESRI going to offer to buy Manifold? And, once bought, will they use the product, raise the price, or bury it in the ground?
Acquisitions are tough, nothing to take for granted. I joined Intel as a result of an acquisition so I have some personal experience of that. The company acquired was extremely tight with Intel with many close personal friendships, mutual corporate admiration and daily contacts working together on projects vital to both companies. It still was very challenging to accomplish the acquisition.
To answer your question, timing is everything in acquisitions and ESRI may have missed this boat even if any of the newer companies were interested in selling.
The time to swallow up a more agile new competitor is after they have developed significantly new technology and are on a technology roll but before they get any market traction or cash flow. Once the competitor gets positive cash flow and profitability and sees their sales taking off it becomes very much more expensive and more difficult to buy them.
For starters, if you are already making more money than you can spend there is no pressure to sell to make more – you tend to be driven by other factors. I grant you that among the sort of competitive people who end up in entrepreneurial companies there’s often a natural human desire to want more, but that declines very rapidly once money ceases to be a limiting factor in the daily life of the company or personnel. Very soon in the evolution of successful companies people switch gears from worrying about cash to thinking about how to do more development more efficiently (ChrisW very astutely raised a key factor that concerns every manager of a large development project), how to invest that development into worthwhile products, how to project the new paradigm that results into a worldwide scale. If an acquisition helps that, they might be interested, but otherwise the fascination with that is so compelling it is very difficult to tempt such people with a bit more of what they have in plenty already, cash.
And if they are motivated by adding another gold coin to the pile, well, if their sales are on an upward trajectory why sell now when by waiting six months the valuation will be greater and you’ll get even more?
That’s especially a factor when new companies play a disruptive role in switching paradigms in legacy businesses (I’m not using “legacy” here as a pejorative, just as the literal meaning of pre-existing types of business). Such markets will often switch very rapidly to a new paradigm once a tipping point is reached. Seemingly overnight, they can go from being totally dominated by a legacy player to a very new paradigm where the formerly dominant player has little or no role.
The most recent example of that was the switch in Internet advertising from portal display ads to paid search, with the former players virtually disappearing as the paradigm shifted to Google’s view. An utterly unknown company that was microscopic in size compared to Microsoft, Yahoo and Alta Vista (remember them?) surfaced overnight and became the leader.
That has happened many times before in situations like word processing (the end of Wang) or computers (the end of minicomputers).
Whether or not you agree any of the newer GIS companies have a chance of pulling off such a paradigm shift, all that matters in analyzing the dynamics of any possible acquisition is to understand that their managers might think they have that chance. If they do, and if they are making plenty of money already, their optimal financial strategy is to “lean into it” and to do their best to push forward disruptive technology to cause at least a partial paradigm shift. Do that and the valuation soars. If you can pull off a real paradigm shift, as the founders of Google discovered you can virtually overnight become billionaires.
ESRI bought Atlas years ago and then buried the technology. But as far as I know (people more familiar with Atlas no doubt can relate the history better), although Atlas was a truly fine product it was not operating anywhere near the technology scale of the strongest modern GIS company, nor did Atlas declare an intention to scale out in the way companies which effect fundamental paradigm shifts tend to do.
The GIS company to which you referred has that characteristic, also shared by prior successes like Dell and Google and Microsoft, of being willing to commit itself with no quarter asked nor given into a really big vision of what needs to get done. In particular, they’ve announced they will not rest until their package goes far beyond what most folks think is the limits of GIS: they have openly stated they want no one with their package to ever have to reach for a CAD package like AutoCAD, graphics arts packages like Illustrator or PhotoShop, remote sensing packages like ENVI and so on. To get all that in addition to the world’s best GIS for an affordable price of $250 is what I mean about thinking in terms of world-class paradigm shifting, and they are actually serious in believing they’ll be able to pull it off. If anything, they are getting more confident as they get further into it. They obviously have a long path ahead of them, but they are making steady progress and will probably be there remarkably soon, possibly even as early as three years out.
Given that intent, the traction already achieved, the surplus of cash and the prospects of even higher valuation either through steady sales or a paradigm shift, even if they were willing to sell to ESRI I doubt that ESRI can afford them. More likely the valuation is already at the level only a Microsoft or Oracle or similar could afford.
In any event, if you have a big vision that you truly believe in and that you know how to turn the crank to get there, and you have plenty of money, there is no reason to be in any hurry to sell yourself to someone else. ESRI might have had a chance, years ago, to fend this threat off by acquisition as they did Atlas. But I think the opportunity for that has passed.
Have to admit Dimitri – one of your best posts I’ve read. It is interesting in the GI market that ESRI hasn’t bought out another player in the market in some time.