In ’94 when I went to college, the Mech Eng department had recently taken delivery of a shipment of DEC Alpha workstations. These were brutally powerful machines; 64M main memory, 3D graphics cards, running OpenVMS, and 6 or maybe 8 of them were available for undergrad use. They were the only machines we had that could run the CAD software I-DEAS, so they were constantly in demand (not least because they were often used by people who would open a single DECTerm and use them as if they were VT100s!). You could run MATLAB or FLUENT or FORTRAN (all software back then was in capital letters) on the older VAXstations, but for anything interesting you would probably need to submit the job to a batch queue and come back later or the next day, whereas the AXPs could do much of this stuff nearly interactively, by which I mean, the time it took to get a coffee or lunch. Unprecedented!
I worked all throughout college, so I had no debt, and in my final year I took out a student loan and spent the lot on a Power Mac, on which I dual-booted MkLinux. 64M RAM, 2×1.2G disks, 132Mhz processor. Holy crap! This thing ran my FORTRAN and C++ as fast as the Alphas! It was fast enough that I could even do numerical work in Tcl! I couldn’t say exactly when the shift took place, it might have been merely weeks before this part of the story, but now ordinary individuals, students even, could own the sort of computational power that hitherto was only available to organizations. The writing was on the wall for DEC, SGI and the rest then. But there were still a few of the puzzle pieces missing.
My first “real” job, in ’98, was at a consulting company, who were paying around £80k/year (or perhaps it was per-quarter, I can’t remember) for a T1-equivalent line, 1.5Mbit/sec. Fast forward to ’03 and I found myself, through a rather complicated arrangement, working out of the offices of Sapient, another consulting company, who at the time were located in a building behind the Old Lady. They had a 512kbit/s line to the Internet shared for the entire office. I had just bought a flat, and gotten that new-fangled ADSL thing installed, at 2Mbit/s down. So at home, with my own money, for my own personal use, I had a more powerful computer than they gave me at work, and more bandwidth too! Something strange was definitely happening, the balance of power between individuals and organizations was tilting alarmingly the other way. Hosting in a real datacentre was still expensive, but nowadays we have EC2, Azure, and many others.
So the means of production are finally in the hands of the workers, yet the corporation shows no sign of going away. But what is it for? Or to put it another way, what is the value of a conventional corporation in hi-tech industries? We’ve seen that it is not that it owns a bunch of “stuff”; anything that a company can provide that a programmer needs to do his or her work can easily be bought; for a few hundred pounds you can have everything you need for industrial-grade software development, including somewhere to run it in Production. It is not even that it has a bunch of source code in Perforce – leave any codebase alone and bit rot will set in, it may go on being a cash cow for a few months, then rivals will clone its features, third party systems it’s connected to will be updated and it will no longer speak to them, regulations or market conditions will change and it will be unable to adapt without constant attention. No, the value of a high-tech organization lies in its ability to identify and solve new and commercially relevant problems, which really means, it has recruited and is able to retain the people who can do that. Its “assets” are free to come and go as we please, and its “capital” is the combination of sum total of our experience and the investment therein to date, and the network effect of having a group of people who mesh well together†.
It occurs to me that such an organization ought to have a radically different ownership and organizational structure than one based around traditional capital. Anyway, this little ramble was inspired by the news that Neuromancer, which I first read in the mid-90s, is going to be made into a film.
† That bit is important; we are social creatures for a start, and many interesting projects are at a scale where an individual cannot realistically undertake them at the speed of the market, if you consider the need to individually subcontract each piece of specialist work, it’s just too much friction‡. So we work for companies, ultimately, only because a critical mass of people we want to work with also work for companies…
‡ I concede that there is value in scale for its own sake; no large organization wants to risk relying on a much smaller organization which is at a higher risk of going out of business. But what if there were no large organizations?