Friday, 19 July 2024 GMT
[ journal | e-cards | reminders | surveys | about ]

What can we learn from ArsDigita?

The Iraq issue: arguments and statements

Information representation: Systasis comments, part 2

The Iraq issue, part 3

Your comments appear at once, but will be censored if they don't follow the usual etiquette guidelines. We do not require registration in order to encourage dialogue, so please don't abuse this courtesy. HTML tags allowed (in the message body only) are B, I and A, hence you should use &lt; instead of <, unless opening the specific tags.
What can we learn from ArsDigita?
Sunday, 25 May 2003 Symeon Charalabides (

Well, fire your MBAs before they bring you down, really. Better yet, don't hire them in the first place, but maybe we should iterate the story first before coming to a conclusion. It's only fair.

I first heard of Philip Greenspun when I read his article "What can we learn from Jakob Nielsen?", in which he summarised the latter's most recent book (then), "Designing Web Usability". It was great to read a programmer's distillation of a book that was rather interesting in principle, though I could hardly bring myself to even imagine paying for a book that would have been, roughly, 60% armchair talk and 40% content, quite ironic given the usual content of Nielsen's banter, and especially since my Kim Wilde collection was not complete at the time.

Philip Greenspun came up with interesting stuff of his own. More importantly, though, he gave all for free. I read his book "Philip and Alex's guide to web publishing" thrice before I could identify/apprehend with it fully and, except for a rather inflated ego and a suspicious psychosis with dogs, he sounded (read) mostly right. Sure, PostegreSQL was only beginning to get recognition in '99 and nobody seemed to be willing to trust a database under GPL. I mean, database? These things are incredibly complex. Surely, nobody can offer such a thing for free with any semblance of reliability when Oracle seizes your bank account for a 'standard installation'? Or not?

It is not polite to digress in such a way. Philip based his ideas on principles, mostly ones found to work. Aesthetics is a matter of taste we can't possibly all agree on (nor would it be fun if we did). Every time I browse I feel a need to fire up Photoshop and distort graphics for sport. Also, he has a considerable amount of this terribly underrated attribute of web-developers, experience, and shares it freely. Most of all, though, his websites work. They work as advertised or expected, fast and hassle-free and I'll take a visually minimal website that works over a fancy & clumsy one any day of this year and the next.

Philip Greenspun founded his own web software company, ArsDigita, together with Jin Choi. Starting off with his own machines and a capital of $10,000 in 1997, the company had grown to an annual revenue of $20 million by 2000. It consisted mostly of hardcore programmers working around the clock to solve the problems their existing customers had or the ones their future customers wanted tackled in order to become such. This process yielded the ArsDigita Community System (ACS), a toolset of modules developed around a central core. The source code was given away from day 1 and directly contributed in making the company well-known, as well as concentrating a development community around it.

Eventually, the need for a substantial infusion of cash became pertinent if the company was going to grow any more. To this end, General Atlantic Partners and Greylock offered $38 million in exchange for 2 out of 7 seats on the board and veto power over a few specific kinds of decisions. Another seat of the board was taken by an outsider CEO proposed by the venture capitalists. That was around March 2002. The company now had an optimal structure and ample venture capital to trudge along.

What happened, however, was exactly the opposite of the intended result: ArsDigita started leaking cash, clients and the respect built over the years right away. The executives' ignorance about software engineering, software itself and their reluctance to admit said ignorance, all coupled with their primary interest in squeezing as much revenue out of the firm as soon as possible led to some terrible administrative decisions:

[a] ArsDigita started charging for the ACS instead of giving it away and making profit out of support.

[b] ArsDigita shipped the last version of ACS way before it was marketworthy.

[c] ArsDigita switched completely from using ACS to a crude and slow Java-based tool.

[d] ArsDigita fired more and more developers, hired more and more managers until all higher positions were manned by people affiliated with the VCs and the original development team (including all co-founders) had been driven away.

Less than two years after the venture capital "boost", ArsDigita had hurriedly been sold for scraps to RedHat just prior to amounting to nothing. It had had no product (and consequently no services) to sell, no market share and no customer base. Philip Greenspun had managed an out-of-court settlement over the law suit filed against him by the VCs using company funds to pay for all litigation expenses. The settlement left him with permanent exile from the company he'd created and a sum of cash that was ill-compared to the company's original potential. The other co-founders had simply been let go. ArsDigita itself never went public.

Here is Philip's detailed account of the relevant events. What had gone wrong?

To start with, software companies are strange. Software engineers are strange, too. When you get down to it, software is a strange thing that doesn't carry traditional values and attributes of goods well. In addition to that, small software firms are even further away from traditional corporate formats than big and established ones. It is wholly unrealistic to expect a professional CEO to take over the chair of such a company and know how to run it. When the people who obviously can't manage their task won't admit it or ask for help, it is high time to get cracking.

There is a severe problem with the average "management mentality". Professional managers rely on taught rather than acquired (or even personal) ethics and they always have something to prove. Analysts have a ball with them. It is not so much that, for the most part, their higher education consists solely of the motto "The purpose of a company is to make a profit". It is that even that is incorrect. A company's purpose isn't to make a profit any more than a human being's purpose is to eat and drink. Profit keeps a company going. Its purpose is to cater to a specific need, to fill the niche in the market it was originally created to. Accomplishing that task gains the company a client base and, subsequently, the funds to keep it going. If there is any money left over, that is profit, but it is a by-product.

It is frequent, and has for some time even been trendy, to confuse cause and effect. When your financial future relies on the distinction, though, caution is advisable. There are no companies who successfully sold only their own attitude to the public. In contrast, there are millions of companies relying on their products and/or services to do their PR for them who thrive even in this day and age. Self respect is important, but self-flattering and flashy ties are not the proper way to get it - it just doesn't feel the same. In contrast, hard work, willingness to learn and uninhibited communication, all under a viable strategy, have been known to go a long way towards delivering the goods for centuries. Can all this be changed by the advent of the "science" of management? Hardly.

Perhaps the most important part in the fall of ArsDigita was the decision to relinghuish control, even partly, in the first place. Now, don't get me wrong: capital is important. Very much so. There's safety in numbers, and this is hardly ever truer than when those numbers are preceeded or followed by a respectable currency denomination. There is also a lot of satisfaction to be had by progressing and although a bank balance cannot express every aspect of a venture, nonetheless it is readily measurable.

When you invite the wolves into the flock you'd better be prepared to take some casualties. Philip Greenspun certainly didn't expect the death toll to rise that high. He just wanted somebody to take over the administrative chores that had become more time-consuming than he cared for, so he could spend more time on the production floor again. He wanted the company to grow further, which could only be achieved that way. He didn't want to lose any clients to the fear that ArsDigita might fold any day, the way so many other software companies had gone bing.

The unavoidable conclusion is, in general terms, that you can't have it all: the more you grow, the less control you will have. In the case of ArsDigita, the VCs managed to bring it down although they never actually controlled the majority of votes on the board. There is an optimum size for everything where the outside friction and the inside pressure cancel each other out and everything just works. Granted, under current economic conditions and especially in the IT race it is mostly "Eat or be eaten", but I can't help feeling that had ArsDigita decided to stay put and be content to never take on software giants, the ACS would have reached version 6 and an unprecedented level of configurability, Philip Greenspun would still be preaching default link colours, Eve Andersson would not have written her bitter recount of the story, and the venture capitalists would have been none the wiser (assuming, of course, that they now are).

As it is, however, ArsDigita became just another example of a software firm that, despite its awesome and proven potential, went bust in the blink of an eyelid owing not so much to bad administration, but rather the very need for administration. There lay an important lesson for all of us, one that is harder to accept than to actually learn. As a last positive note (for oh such romantics are we) it should be mentioned that the ACS lives on as OpenACS, and now stands for Open Architecture Management System. Since it can run on PostgreSQL as well as Oracle and AOLserver is now free, it is a complete open-source product and distributed under GPL. A tribute to the hard work and vision of the original developers and a real-life useful toolkit, it is certainly worth spending a little, or more, time toying around with.

 Send this article
Printer-friendly form
Your e-mail: remember
Recipient's e-mail: remember
 Comments - [ write ]
  IP: OS:   Agent:     Overhead: 0.003 sec