Coasting on Empty

Background: I wrote the first draft of this rant in response to someone who questioned my refusal to release Emblem (and most of the software I build using it) under the GNU GPL.

What I'm doing, unpaid for, while unemployed despite what must be over 100 job applications (I've lost count), is research and development, the sort of thing that industrial research labs used to do. Understandably, I don't want the companies who should be paying for R&D to get the fruits of it for free. They can pay me if they want it, but if they don't, the can't have it. That's why Emblem's on a shared source licence, instead of the usual GPL (which I'm happy to use for non-innovative stuff), and a similar licence will apply to most of the software I build using Emblem.


I think the industrial landscape has changed since the mid 1980s when Stallman started GNU. Remember that he didn't do it out of the goodness of his heart, or to gain market share. Anger drove him -- he did it to get back at Symbolics.

Back then, there was serious research being done by industry (indeed, Stallman's least favourite company spent 20% of its budget on R&D). Now, there's close to zero. Worse, I have suspicions that in the UK companies are claiming to the Inland Revenue that they're spending money on R&D which is going into straight development, and pocketing the tax break. BT, one of the few companies which still does R&D, has cut its R&D workforce by 90%. Meanwhile, universities are taking up industry's slack (often industry funded because PhDs are cheaper than research staff) and doing the applied research for them, so often they don't have time for pure research.

So the stream of new ideas in software is drying up, and we get old ones from as long ag as the 1970s, recycled. We're coasting on empty. There's still progress in hardware, but that's being made by a tiny number of people. In software, there's just The Illusion of Progress.

I'm aware that most of the job ads of the 1980s asked for CICS+COBOL+JCL or FORTRAN and that software was done then by the waterfall methodology (as it often still is) and that many places back then still used punched cards, and to people from those backgrounds there has been an improvement. We're now using systems eerily like the Smalltalk systems developed at Xerox PARC in the 1970s (see http://www.digibarn.com/collections/systems/xerox-8010/index.html): just cheaper, faster, with bigger memories, in smaller boxes, but without Smalltalk -- so we've relied on hardware advances but regressed software-wise.

If you look at typical "GUIs", they're just slick IBM 3270 emulators. There isn't much in the way of graphics outside of PlayStations. And even the games manufacturers are short on ideas.

It took over 1000 years for things to get better technologically after the fall of Rome, and that was only possible because of ideas from the Islamic world, i.e. from outsiders. Now, with globalization, there are no outsiders. Even the Japanese are using Windows NT, Java and XML. So the worst case scenario is that things will never get better. Admitted that's highly unlikely, and that I'm missing something that might cause technological progress to resume, but all I see is an emphasis on teamwork at the expense of the individual (see The Organization Man), an emphasis on short-termism across the board (see Short-termism and the Death of the Future), and the absence of a balance of power now that the Warsaw Pact has collapsed and Japan is not the economic power it once was. So what is it that I'm missing?

Through a Glass Darkly extrapolates into the future, and makes some recommendations.


Some background reading:


There were still a few exciting developments

which are happening in academia, and as open source efforts.

This page was linked to from

  • index.html
  • and was last updated on 2006-01-22 at 22:14.

    © Copyright Donald Fisk 2006