Over on his ZDnet column Universal Desktop Ryan Stewart, who describes himself as a Rich Internet Application mountaineer, makes his annual predictions about where the Web is going. Prediction number seven caught my eye:
7. The days of smaller RIA technologies are numbered. I hate to say it but I think technologies like OpenLaszlo and Curl will continue to gain traction in some niches but won’t see widespread adoption. Those companies will still bring revenue but Microsoft and Adobe are pushing too hard and putting too many features into their runtimes for the smaller companies to keep up.
I hate to say it but I think he’s right1.
Watching RIA trends play out is a bit like watching a Film Noir movie (DOA comes to mind), where the good guys don’t win and the bad guys prosper– but not because of any particular genius on their part, merely because of inexorable fate.
I observe this particular web-noir movie from the perspective of an extra actually on the screen. I play a guy cut down by a stray bullet for the crime of being in the wrong place at the wrong time. For I was manager of information architecture at Curl from April 2000 until being laid off along with most of the engineering staff in April 2002, and I was the sole doc guy on the OpenLaszlo project from April 2003 until being laid off November 2007 as Laszlo Systems gave a quarter of its staff the axe.
Below the fold, a few brief observations on the Web 2.0 drowning pool. Said observations are undoubtedly greatly corrupted by time and rationalization, so take them with whatever quantity of salt you like. I’m just recording them for my own record.
(1)Note: although Stewart now works for Adobe, he’s always been fair-minded about competing technologies. I don’t see him as a shill.
UPDATE: edited a few sentences for clarity & one new comment at the end, in response to private reactions to this entry.
Curl: A technology too soon, a management too lame, a Kruger too Freddy
Curl Corporation provides a runtime (called “Surge”) that runs in a browser (like Adobe’s Flash Player), a compiler, and a development environment.
When I joined Curl in the spring of 2000, developers there were working on apps that were more advanced than anything you can do with Ajax (or Flex, OpenLaszlo, etc) today. Curl had (& still has) a really cool 3-D API, good sound integration, an elegant way of handling text, all predicated on a nifty just-in-time compiler. You could scarf XML data off the net and use it to drive a 3D animation with sound, and render text in any form you wanted, whether latin, Kanji, Korean, or whatever. Incredibly powerful stuff. Years, perhaps a decade, ahead of its time, technology wise.
So why did the company fail?2
Because the runtime failed to catch on, that’s why. You cannot run a Curl program unless you download and install the Surge plug-in, and too few people were willing to do that. This presented a classic chicken-egg problem: because there was no installed base, there were few application developers, which meant there were few applications, which meant that nobody had much incentive to download the Surge player, etc. So there was no way for us to make money. Every new technology faces this problem, of course, yet many succeed by creating a groundswell of enthusiasts among developers. Why was there no Curl developer groundswell?
Among many other reasons, I can cite three:
1) Curl had a wacky-nutso licensing scheme whereby customers were to be charged per byte of Curl code that their customers executed (it sounds baroque because it is). This meant that bytes had to be metered, and that user machines had to send information to Curl, Inc, in order that we could bill our customers. Turns out that the very concept of “bytes executed” is ontologically complicated, for reasons that I’ll spare you. So engineering-wise, the revenue model was very tricky. Privacy-wise, it was very suspect. Public relations-wise, it was a disaster.
2) No Mac, No linux == no developers. The fact that the runtime was only fully supported on on Wintel machines limited its appeal to developers.
3) The language itself is very heavily influenced by the MIT cult of LISP (scheme, etc). To your typical Java, C++ or even JavaScript developer it looks & feels very foreign. Developers took one look and moved on.
So, there was no constituency helping to make our case for us.
Basically, although we didn’t realize it at the time, we were in a fight with Macromedia (now Adobe) about whose browser runtime would own the day. Back in those days there were lots of competing runtimes, Java on the client was still considered viable, and people would often download several plugins to try them out. One of the challenges facing the company was to get developers to download the runtime so they could see how cool it was, but this was a problem faced by all Curl’s competition–most of whom are long gone.
But every time you download a runtime (plugin or web-aware program like Kazaa or RealPlayer), you incur a security risk. I myself got bit in the ass enough times that I adopted a blanket policy of “no downloads”, (to the amusement of my more hardcore colleagues, who seemed to welcome downloads as a challenge, like bullfighting or something.) My attitude was, Download an executable? You must be shitting me. I wasn’t the only one who adopted that stance. Eventually the marketplace decided that it only needed one plugin, and that plugin was the Flash player. From there it followed that Curl was fucked as an Internet company. (And indeed, there were many funny, if painful, discussions about Curl on the old FuckedCompany site.) It didn’t matter that Curl’s plugin was technically superior to the Flash Player. Betamax is technically superior to VHS. So what? By 2002 it was already clear that the Flash Player was pulling ahead of Java, not to mention all the other also-rans.
But if Curl’s technology was that good (and it was), why did it lose out to Macromedia, a company who were, 8 years ago, a purveyor of shrinkwrapped tools best known to web developers for making crappy banner advertisements? Remember, in 2000, the Flash Player version 2 was an underpowered animation engine, not the powerful RIA platform that Flash Player version 9 is today.
The main answer is that Curl’s senior management sucked. The founding president was doing God-knows-what. He never came out of his office, so it’s really impossible for any of us to know what he was doing. But he certainly wasn’t leading us, the employees. The VP of Engineering was a classic yes-man who prided himself on hitting schedules (he had three clocks in his office, set to atomic clock time), but who had no insight into what we should have been building in the first place. He was a “Charge of the Light Brigade” type who believed that his job was to follow orders & convince his subordinates (I was one of 5 engineering managers reporting to him) to follow orders. The fact that the orders were stupid concerned him not in the least. The VP of Marketing was a very distracted snake-oil salesman with grandiose visions of taking over the web by an insane licensing scheme. The chief technology officer was a nice guy with a classic case of academic arrogance: he once told me that he didn’t use the Internet. That is literally true: the CTO of a company that had set for itself the goal of reinventing the Web did not in fact use the web. He did not look at websites. He didn’t read Slashdot or shop at Amazon. To him the Web was an abstraction, a computer science problem. That is to say, he figured that he could come up with a solution to the Internet’s problems in a thought experiment, much as Einstein had come up with the Theory of Relativity by expedient of a thought experiment. Coming to Curl from a professorship at MIT, this CTO had had no experience with customers whatsoever.
And, Curl had a relative pantload of investor money, which ended up doing as much harm as good. They spent if not quite drunken-sailorly, then at least liberally, to hire a lot of very, very talented software engineers–without having a very good plan for how to use them. The corporation had a great technology with few compelling applications that showed its real power. So, a whole passel of “applications engineers” were instructed to “invent something.” Not surprisingly, they foundered. I remember going out to lunch wih some new colleagues on my first day on the job. Curl was a startup, so I was expecting that things would be fluid. But when I asked how the company made money and what everybody was working on, and everybody wanted to hear my ideas, I sensed I was in trouble.
One result of the absence of clear leadership from anybody on the senior management team was that there were too many worker bees with an unclear sense of what they should be doing with their time. Fiefdoms arose. Work was duplicated. Office politics became nasty. Morale went into the toilet.
The money ran out eventually. The venture boys on the board fired the useless president & found a new monkey to run the show; he brought in a bunch of young cocky mercenaries, and they began decimating the place, Tony Soprano bust-out style. The fourteen-person group I had been managing (which included wetmechanics Gary Gray and Howard Stearns) was reduced to three people. I personally gave the pink slip to most of the people in my group; Howard and I managed to hang on for another few months under the strategic protection of the one member of the Board of Directors who had personal integrity and a brain in his head3, to write a strategy paper that went nowhere. Then the axe fell on us.
Anyway, blah, blah, blah, this is ancient history, biased, unfair, mostly forgotten even by me, and who cares anyway.
And yet, and yet, Curl is still around, with several of the engineers who were old-timers when I arrived nearly eight years ago still there plugging away. I would kind of like to see them succeed, just for old time’s sake. But how can they even still be around, six years after their spectacular dot-com-bust meltdown? Why is Curl.com still up?
Turns out that some Japanese company bought up the rights to the company in 2003 or so, and kept on board the core of the remaining engineering staff–about twelve people. They abandoned any pretense of becoming an Internet company and set out to be an “enterprise” intranet vendor, focusing on Japan and Asia. Where, evidently, they’ve had some real success. I spent years expecting to hear through the grapevine that they had finally thrown in the towel and closed their doors, but I no longer expect their immanent demise. Perhaps an indefinite attenuation. . .
And now I see that InfoWorld has given Curl its Technology of the Year in the Rich Internet Application category.
Truly, they may be unkillable, like Freddy Kruger. It will be interesting to see what Ryan Stewart has to say about them this time next year.
Laszlo and OpenLaszlo: Genius is Pain
I spent four and a half years at Laszlo Systems and I still have a lot of friends who work there. So my take on Laszlo and OpenLaszlo is less detached than my take on Curl. Plus, in order to get my measly severance check I signed an agreement in which I promised to not bad-mouth the company. Not that I would want to, mind you. But also, I can’t. So I don’t have a lot to say.
But briefly, it’s hard for me to not think that the air is going out of the OpenLaszlo balloon, and if that happens, I don’t see how Laszlo Systems, Inc, can survive. Which is a crying shame.
Like Curl, OpenLaszlo is an RIA technology. Unlike Curl, its programming model (Its language, called “LZX”, is XML + JavaScript which compiles to either Flash or DHTML (or, presumably some day, to Java, Silverlight, etc)) will be familiar to most web developers. It’s an elegant, nifty model. If you want to find out more about the technology, check out the OpenLaszlo.org website. I wrote most of what’s there, including the technology white paper and virtually all of the technical documentation. If you’re at all technical, at least check out Laszlo in Ten Minutes to get a feel for it. It’s cool.
The genius of Curl’s founder and Chief Technology Officer David Temkin was to realize that although the Flash Player was a pretty darn good Internet application program by the time Flash 5 came along, the Flash movie/Actionscript programming model sucked. David and Oliver Steele and a bunch of other cool and extremely smart people, many of whom still work at Laszlo, came up with a programming technology that would exploit the Flash player’s ubiquity–thereby completely sidestepping Curl’s single biggest obstacle–while making the process of developing web apps not totally suck. In fact, their language is so powerful that it became possible to create whole new classes of non-sucky web apps.
Macromedia, which became Adobe, liked David’s approach so much that they ripped it off, calling their version MXML, which evolved into Flex.
Because Macromedia/Adobe, Inc, is a bajillion times bigger and richer and better-known than Laszlo, once they began marketing their Laszlo knock-off in earnest, Laszlo realized they couldn’t go head-to-head. Their strategic response was to open-source their technology. The “Laszlo Presentation Server” became “OpenLaszlo” (Sarah Allen wrote a nice essay about what the “going open” experience was like to us engineers on the project; links to her blog & my comments on it are here.).
But even with its platform technology completely free and open-sourced, Laszlo still could not take on Macromedia if it were merely a “me too” vendor of XML/JavaScript for Flash-based applications. So, opensourcing was a good start, but it was not enough to establish Laszlo in the marketplace. But! Unlike Macromedia, Laszlo is not wedded to the Flash platform. The core technology was designed from the start to make it “easy” (easy for Leonardo, perhaps, as Dylan Thomas might say, I digress) to compile to other runtimes. So, Laszlo undertook an ambitious project, called “Legals”, to create an Ajax compiler for LZX. The significance of legals is that it really does bring web developers much, much closer to the old Bill Joy/Java holy grail “write once, run everywhere”.
This argument is much easier to follow in the OpenLaszlo white paper, so if you care enough, go read that.
It’s pretty fucking brilliant, actually. (The OpenLaszlo strategy, not my white paper about it, although that’s not bad. . .)
I wrote about the whole saga of the Legals project, the most interesting software project I’ve worked on since 1980,here.
Legals should have been a smashing success, but it has mostly fizzled. Why?
Here’s the thing: Laszlo was (and is) underfunded. Whereas the greedy venture bastards who funded Curl drowned the baby with too much cash and not enough strategic vision, Laszlo’s greedy venture bastards starved it with too little money to reap the killing that OpenLaszlo should have brought them. Here’s what I mean.
Ever since making the fundamental technology open source, Laszlo has had two teams internally, and two personalities. One team makes stuff for sale (either work-for-hire or off-the-shelf), and one team makes the OpenLaszlo platform, which it gives away for free. Although there is (was) $$ from investors, the company was straining to get cash-flow positive, which meant that the revenue-generating parts of the company were funding the open-source parts of the company. This was too big a burden to put on the backs of the revenue-generating part of the company.
During the entire Legals project, which lasted, if we’re going to be honest here, two years (not the one year it officially took), the entire OpenLaszlo team was focused on creating a new runtime and new infrastructure for future runtimes. Which is to say, we were NOT adding new features to the existing runtime. We spent two years running in place, albeit running in place in two places at once, which is quite a trick.
What did Ryan Stewart say? Microsoft and Adobe are pushing too hard and putting too many features into their runtimes for the smaller companies to keep up.
Sad to say, he’s exactly right. That’s exactly what happened.
It’s probably too late now, but it didn’t have to be this way. Unlike Curl, which was rudderless for at least the two years that I was there, and evidently before and after Sundman as well, Laszlo has always had strategic vision (embodied by Temkin, and others), and it’s always had customer focus. With never enough money in the bank, the company is nimble; its engineers are grounded in the real world. Through most of its history, Laszlo Systems, Inc, has had competent, if not always god-like senior management. What it hasn’t had is sufficient money to build-out its vision. On the OpenLaszlo team, we had to choose between three equally important tasks: attending to and “growing” the worldwide OpenLaszlo community, supporting multiple runtimes, and adding new features. We only had sufficient engineers to do one of those tasks well, although we tried to do all three. And now it appears that Laszlo’s being overtaken by a wave it was brave enough to ride first.
I remember taking two walks down city streets with David Temkin. The first one was in the summer of 2004, in Boston. David had come out from California to meet with Oliver and me and two other members of theLPS, soon to become OpenLaszlo team. The subject was whether to go open. We went out to lunch at a bistro in the South End on a beautiful sunny day, and on the walk back we hashed it out: the pros of going open, the cons, the challenges, the work to be done, how the project would be organized. As happened many times during my years with that team, I felt lucky just to work with people who were so smart, and also nice.
The second walk was a few years later, in San Francisco. David treated me to dinner; that was the night I finally met his wife Stephanie. We dined at some very nice, trendy place South of Market, and I pounded back a ton of girly rum drinks.
On the way from David’s car to the restaurant, we were talking about youtube or some other company, I don’t really remember, some infant, brand-new company that had more of a gimmick than a real product or technology, and which had nevertheless just been acquired for some obscene amount of money. I remember David making a dumb show of pretending to kick over a trash can and smash a window in mock frustration (I guess you would have to know funny, foul-mouthed and cynical David can be to get the full picture. . .). Why did these yahoos strike it rich while Laszlo, which was full of geniuses, and which had real, valuable technology, not a gimmick, continue to struggle? Indeed, genius is pain sometimes.
In a fair world, OpenLaszlo would be the hottest open source technology around, I’d still be working there, and all us good guys would be rich.
Oh well. In film noir the good guys never prosper.
2) Or has it failed?
3) He resigned from the board soon thereafter.
UPDATE: Final comment. To be clear, I don’t fault anybody for Laszlo’s current predicament. As I said in the intro, it’s likely just fate. But lots and lots of people use OpenLaszlo; more than use Curl, I would guess, by a few orders of magnitude. OpenLaszlo’s a good thing, a quality bit of software, but it probably won’t catch on at an exponential rate, and if it doesn’t grow exponentially it will fade away. Take a look at that video I linked to, above, of the guy on the giant wave. That’s a quality ride, and a thing of sublime beauty. But he can’t outrun the wave.
I don’t understand the attitude of the investors who have put virtual nickels and dimes into this enterprise. OpenLaszlo could have been, should have been, (should still be?), the lingua franca of the web. Although Laszlo Systems, Inc, loses money on OpenLaszlo, if OpenLaszlo took off, that would allow Laszlo itself to become another Adobe, if not another Microsoft. And yet Laszlo management has spent all kinds of time groveling for investments of a few million bucks. Call me stupid, but if I were sitting on a venture fund of a billion bucks, and I were into Laszlo for five million bucks, and Laszlo came knocking looking for another five million bucks to really capitalize on their vision, I would put in thirty million. Laszlo’s already proved that they have the vision and the ability to deliver. “For want of a nail the shoe was lost, for want of a shoe the horse was lost, for want of a horse the rider was lost, for want of a rider the battle was lost. And all for want of a nail.”
But I’m not a billionaire Silicon Valley investor, I’m just an unemployed tech writer. So maybe I don’t know nuffin after all.