Through a Glass Darkly

In 1973, Christopher Hutt's book The Death of the English Pub was published. At the time, most of the British brewing industry was in the hands of seven companies, six of whom (the exception was Guinness) were getting away with lowering the quality and strength of their beer, taking over smaller breweries that produced good beer and stopping its production and replacing tenant-run pubs with management run pubs, or closing them entirely. Traditional cask-conditioned ales were under threat everywhere, and were increasingly hard to find, in the face of heavily marketed, bland, lifeless, standardized, industrially produced keg beers.

The future looked grim. Hutt's dystopian vision, an extrapolation of then current trends, is contained in the last chapter: automated theme bars serving industrially produced beer with no flavour (unless fruit- flavoured).

Happily, things never materialized in the way Hutt feared they would, in large part because Hutt himself alerted people to what was happening, and also because of a relentless campaign by CAMRA, and legislation introduced by the Tories during the 1980s which required tied pubs to sell a guest beer, which had to be cask-conditioned. (The Thatcher government intervening in the free market? Yup!) England is again a beer-lover's paradise, and beers such as Watney's Red Barrel, Double Diamond and Younger's Tartan Bitter are distant (though still unpleasant) memories.

What does all this have to do with software?

Today's software industry is in an even worse state.

Today's software industry is in a worse state than the brewing industry was in when The Death of the English Pub was published. We're down to two operating systems (Windows and Unix) with the same user interface, and a handful of languages (led by Visual Basic and Java) optimized for mediocre Commodity Programmers following an industrial process.

Progress was steady for the first 50 years.

The first computer was built by Konrad Zuse in 1941. Progress was steady until software technology reached its zenith in the late 1980s. Back then artificial intelligence was big business, with expert systems in fairly widespread use; advanced programming languages like Lisp, Prolog and Smalltalk were in widespread use in industry; and Virtual Reality was a hot technology.

We've been in reverse gear since the end of the 1980s.

Since then, we've been backsliding. The World Wide Web is old technology -- the first distributed hypertext system (Info) was up and running on ITS machines in the mid 1970s; by the mid 1980s, Symbolics made hypertext look sexy in its Document Examiner. Apart from the World Wide Web, nothing much else of interest has happened.

Take a look at this screenshot from the Xerox 6085. Look carefully at the date: 10-29-80. Almost 25 years on, and we have much the same thing in colour. Things got better during the 1980s, only for it all to unravel in the 1990s. If this recent trend continues indefinitely, by 2040 all computer systems will have ceased functioning.

Could it happen? Highly talented software craftsmen and computer scientists are being marginalized throughout the software industry. Their understanding of the craft is no longer valued by a management which has engineering envy, and wants to run the industry like an assembly line. This process will continue for the foreseeable future.

Software is becoming more bloated, and buggier. It will be fixed increasingly by work-arounds by less competent programmers who no longer have a clear understanding of how the system works underneath. Virus writers will increasingly exploit bugs in system code.

Software patents, which now seem inevitable in Europe as well as the United States, will make it impossible for small independent software developers to develop software, as they are litigated out of existence by companies with large patent portfolios.

What I've done is, of course, just a rough extrapolation of current trends a long way into the future. It's a warning of what could happen, not a prediction of what will. I'm doing what Christopher Hutt did, not what Mystic Meg or the Gartner Group does. (If they were correct, knowledge based systems would be a multi-billion dollar business today and I'd be on the gravy train for life, not an unemployed computer scientist.)

(Chris Winter, once a cow orker of mine, nearly came to grief over a similarly far-out extrapolation, Soul Catcher, which at the time was misreported in the same way by everyone in the press except Fortean Times and the now defunct Scallywag who were both reasonably accurate. So if you quote me please cite this page.)

It's interesting to look at alternative futures, e.g. Robotic Nation. The problem here is that it is based on the assumption that progress in AI has continued at pre 1990 rates. It clearly hasn't: the last major breakthrough was subsumption architectures in the late 1980s. So relax. A few jobs might be automated out of existence using current technology, but that's about it.

What can be done?

There are plenty of others like me. We're all marginalized in one way or another -- either out of work, in humdrum jobs not fulfilling our potential, or running businesses which are struggling because prospective customers don't believe what we're capable of.

First, we should try to understand Why This Is Happening. Then, do something about it.

Perhaps a guild of software craftsmen (and women) could be set up. Instead of a certification programme based on rote learning and examination, qualification should be by authorship of a masterpiece, a significant software system written by the candidate alone on an open or shared source licence.

Public relations might work. An example: thanks to a public relations campaign, the Tory government became convinced of the economic value of Scottish Gaelic and massively increased the funding available for Gaelic television. Gaelic, a language with minimal official status, is spoken by at most 60000 people, all of whom are fluent in English, and the Tories were hard-headed free market capitalists. Impossible? Yes, but it actually happened.

Perhaps the tax credits given to UK companies for research should be revoked, and replaced by a tax imposed on all companies, and refunded to those companies which produce evidence of research actually having been done and published. The tax should be hypothecated, and provided as funding for independent researchers.

Lastly, I'll throw my voice behind the many others who have already made the point that software patents will stifle innovation rather than encourage it.

This page was linked to from

and was last updated on 2006-01-03 at 20:56.

© Copyright Donald Fisk 2006