In the old horror movies, no matter how often the villagers clubbed together to drive the Monster into a burning building or over a cliff, or to a gelatinous end in some b&w swamp, somehow you always knew the next Friday late-late show would have the lumpy fellow reborn (or re-electrified) anew, once more to wreak terror upon the land through, tragically, just trying to communicate.
By Gary Flood
This image seems appropriate talking to the new folks at the former ParcPlace-Digitalk Inc, of Sunnyvale, California; for it seems the $45m (in stock) July 1995 merger of Smalltalk powerhouses ParcPlace Systems Inc and Digitalk Inc was in fact as unholy and ultimately dangerous an act of constructive surgery as anything Victor Frankenstein did, at least to their investors; and just like in the old Karloff sequels, along have come a new team of eager re-animators, convinced this time they have the right therapy to make the revived old boy behave himself. Our tale begins in 1987 when Xerox’s legendary Palo Alto Research Center (Parc) helped alumnus Adele Goldberg set up a company, ParcPlace, to exploit her work in the pure-play object-oriented programming language Smalltalk. After burning a fair amount of venture capital the company grew to the point it could mount a well-fancied initial public offering in 1994, with shares hitting $20 and more. That year sales more than doubled; in 1994-95 sales rose 47% to $39.1m, with net profit climbing 93% to $3.8m.
Flush with success
Smalltalk’s hour had come, it seemed, with ParcPlace reaping the most benefit – 35 Fortune 500 companies had each spent at least $100,000 on its server-based VisualWorks Smalltalk programming environment. Flush with success, we were told, the company’s wise heads decided to merge with more compact and privately held desktop-centric Digitalk. The combined firms claimed they were now the world’s largest independent objects company with $55m sales in 1995. But how quickly that period when the Monster was still playing with children turned to him rampaging around the forests. In 1996’s fiscal sales touched $49.6m, but on losses of $10.4m (merger costs accounting for $5.1m). In fiscal 1997 (the company’s year end is at the end of the first calendar year quarter), that loss had snowballed to $22.5m on turnover down a quarter, to $37.3m. Desperate measures – closing of its Austin, Texas office, cutting headcount from 350 to 190 – didn’t stop a meltdown in its first quarter of fiscal 1998, with revenue dropping 56% to $5.1m, with another whopping loss, this time $2.5m, with the company recording an astonishing 75% drop in license revenue year on year. Neither did the extraordinary public letter on its web site in February (CI No 3,092), where the company bemoaned the fact that while the market had caught up with objects Smalltalk had had its day, and even though Java was essentially an inferior product it was still the mindshare leader. The company’s share price is now barely a pulse above $1, and has been below it at times; the company’s senior management have all been tossed out of their jobs and into dungeons; and that’s all she wrote.
Language for eggheads
The verdict of history on ParcPlace will presumably be that Smalltalk was a language for eggheads and it was all a waste of time, like Lisp, the prehistoric AI language. But that would be too simple, according to Jim Smith, since August 7th vice president of worldwide sales and marketing, and one of the two main new faces at the company recruited by new CEO Gene Coda, the other being the new CFO Ron Clear. For he claims ParcPlace- Digitalk’s problems came just as much from the fact that it was ParcPlace and Digitalk, and not from the Java phenom. ParcPlace did not merge with Digitalk, it took it over to try and forestall an unnamed third party (not IBM) entering the Smalltalk corporate market, and the cultures never blended. ParcPlace, probably due to its roots in Parc, never stopped being a research company (with very little of the ‘D’ in ‘R&D’ ever being done, he sniffs); Digitalk was much more customer (and NT) focused; and to pacify egos in the competing camps a sprawling number of overly ambitious (the Jigsaw endeavor to unify the two product families) or irrelevant (an intranet division, some dabbling into banking object frameworks) or wasteful (at one time there were five separate US based engineering facilities) follies were tolerated by a management paralyzed by its prior, probably historically unrepeatable, success.
Meanwhile the Java thing did happen, but in essence, according to Smith, we did it to ourselves, citing in addition the PR nightmare of that open letter, or in his pithier version, the suicide note. So if the prior team screwed up, what chance is there of successfully sticking the life juice back in the scarred cadaver? Lots, says the new company, and since this week it will become a new company in name, too, becoming ObjectShare Inc (in honor of its July 1996 acquisition of 15-employee strong Smalltalk house ObjectShare Systems Inc), it deserves a listening. Smith – previously chief operating officer at Select Software Tools Inc – and Coda have identified the way forward as not being identified as a (Smalltalk) product company, but as an OO design and services company primarily. From now on the company’s customers will see the consulting face first, not the product salesman. The company is also now ecumenical, pitching Smalltalk but being able to catch Java, in that it is leveraging its experience in OO concepts in general acquired through all those years of building Smalltalk systems, but is more than willing to implement in Java, anything, Swedish if the customer wants. VP European Operations Russell Prince-Wright offers the example of something he has been selling in Europe, Goods, or Guaranteed Object Oriented Development Strategy, as a model; this package combines training, services and software, in not just Smalltalk, even Borland International Inc’s Delphi client/server development tool, and is likely to be much more the kind of thing ObjectShare will offer long term than Jigsaw in the future.
Smith points to the 2,500 Oracle Corp consultants who are being trained in OO concepts by ObjectShare as a harbinger of this new wave. So ObjectShare will be a training company? No, we will be a product company led by a services led strategy, Smith answers, and in addition, one that will be able to make enough money this way until Java, currently in his view a $50m market at most, becomes a true enterprise arena, probably in late 1998. Or even if it doesn’t – the added value is in being able to help customers build distributed object systems, not being able to crank out the most lines of Smalltalk or Java or even C++. Coda and Smith are frantically busy rushing around talking to customers, hiring the Geoffrey Moore-influenced Chasm Group change management specialists to evangelize their own staffers, though there are now 40 less of those folks, preparing to sell off a few more non-core parts of the company and technology, and generally making a fairly convincing argument that they can turn the company around in one quarter. We have made a lot of our customers unnecessarily frightened and nervous, especially over the past year, he notes. We’re asking them to give us 90 days to see if we can turn this around. Given the pain of a transition to [IBM’s VisualAge for Smalltalk], it’s surely worth waiting that long. The end result will be that the days of those 50% growth rates are gone and will never, never return, but that with around $15m total in the bank we’re in not a bad position for a start-up that will get back to the kind of rates of return you see in mature marketplaces, more like 20%.
Sensible acquisition target
Will ObjectShare’s brutalized share price ever get back to plus- cash value trading? Smith is sure it will, but he is open to discussing the possibility that instead the company will simply recover to the point that it could be a sensible acquisition target for a Rational, even an Oracle (VisualWorks as Sedona? Think about it). Ironic given that a bungled takeover was a big factor in the company’s collapse – but Smith says what he ought to say, that if that makes sense for the investors, that’s what will happen, but the new team wants to see if this idea works first. If this particular Frankenstein Monster ever gets off the slab, it’s likely to be a much more sheepish and open-minded beast than before, and that’s never a bad thing in the software business. The acid test comes with the next set of figures, due late October, when we can see how convinced old, and new, customers are in the Baron’s latest creation.