ArsDigita

A Cambridge, Massachusetts based web consulting firm founded by MIT computer science professor Philip Greenspun. ArsDigita (literally "way of the finger") claims that its empoyees have an average education level equivilient to a Master's degree in computer science from MIT. Ars digita sells tcl/tk based community systems with an Oracle and Netscape server back-end at around one million dollars each. Their projects are essentially implementations of the ArsDigita Community System, an Open Source set of tools they distibute for free. The toolkit is based on Greenspun's well-known photo.net, among the oldest major web-based community systems (founded in 1995). An example of one of their cooler sites is scorecard.org, an EPA pollution report. There is a yellow Ferrari parked in the firm's Central Square parking lot which will go to the first employee to recruit ten new hires.

Arsdigita is dead. That should probably be noted here.

Blame-free chronology:

1. Late 1999: Ben Adida & Don Baccus, interested in developing sites without paying Oracle 5 figures, fork the ACS to use Postgresql, later renamed OpenACS. Many developers follow, for reasons both financial (above) and personal: aD was increasingly disinterested in outside developers and rebuffed outside patches and modules. Severe not-invented-here reek.
2. March 2000: aD forms a toolkit team to develop the toolkit without client-appeasing distractions. Initial plan to refactor and debug the patches and enhancements drifting in from client sites is soon scrapped in favor of a total rewrite. Total estimated developer size: 80; est. toolkit team size: 6-10.
3. Concurrently, VC funding deal is signed, to the tune of $38 million. Note that this is $38 million to a company doing $20 million per year revenue, with presumably high margins and no debt. Warning bells should be ringing right about now.
4. April 2000: hired-gun CEO is found and installed. Leadership now enters poorly-defined phase.
5. Summer 2000: ACS rewrite rewritten again -- this time to be done in Java.
6. Late 2000 to early 2001: aD deprecates the working ACS3 for the still-in-development ACS4, basically choking revenue; aD grows to 200 employees; ACS4 rewrite team grows to 50 people and software becomes totally incomprehensible; OpenACS passes aD in terms of visible development activity.
7. Concurrently, the founders and VCs are at each other's throats; VCs make some stunningly bad financial moves, and founders make some stunningly bad social moves (one fave being Philip writing a company-wide email comparing software development with the Holocaust in a fantasically strained analogy); VCs raise the tiers of management, adding more executive employees who don't code, sell, or support.
8. Again concurrently, the VC management works some social re-engineering: developers are trained to lie to customers, clients are pushed away from developers with increasing layers of salesdroids, aD works to push away the outside developer community before closing the source. aD refuses a partnership with Microsoft.
9. Spring 2001: VCs push 4 out of 5 founders out the door. Philip sues to reclaim control with a suit asking whether shareholder rights (he held >60%) trump investor rights. VCs settled with Philip with a tasty cash deal for his silence. Exit Philip, stage left. Mass layoffs begin.
10. Rest of 2001: aD 'releases' ACS4.2. Everyone, especially the developers, admit that it is pre-alpha, even after an almost total refactoring over 4.0.
11. Late in 2001: ACS4.6, released under a modified source license that possibly excludes OpenACS participation. Not very popular because of that, and thus 90% of the user base doesn't touch it. "Look for a raft, we are now sinking."
12. Layoffs continue; aD cash reserves bottom out.
13. February 2002: aD sold to RedHat for pennies, almost everyone remaining laid off, including the developers. Note that they went 2 years without a stable release. Predominant theory for sale is that RedHat and aD share VCs, so VCs forced the sale so they could put 'sold aD' instead of 'destroyed aD' on their resumes.

There are a number of different viewpoints on this Spectacle; you can find Eve Andersson's, Lars Pind, and many others on their website, Philip's viewpoint can be pulled out of web.archive.org, the fuckedcompany board on aD's sale, and the VC's lined pockets. None of them are totally true, which is good to keep in mind as this little play was so ... varied. The founder's perspectives are fun if you enjoy white-washed moralizing, but the FC boards have all the entertaining little bits of how horrible human beings are to each other. Fun stuff; I'm waiting for the movie.

So far the only lesson I've taken away from the arsDigita failure is that you *can* make serious money by coding and releasing open source software -- provided that you don't refuse the unpaid involvement of other coders, that you aren't an egomaniac and possibly sociopathic, that you don't wipe out everything with a disasterous total source re-write, and that you don't let greed talk you into getting into bed with VCs who will destroy every unique, profitable thing about your company in their attempt to make it fit into their worldview.