Off the Beat: Bruce Byfield's Blog

Outsiders often criticize free software because it offers too much choice. Choice confuses people, they say, and free software would be more efficient if everyone concentrated on improving the best application in each category instead of developing alternatives. To me, this argument has always seemed conditioned by monopoly, but recently I found reason to believe that it couldn't be more wrong.

The insight came when I was trying to set up a local Apache installation so I could develop a Joomla! website, using MySQL as a database. I am not proficient in MySQL, and, in a moment's carelessness, I wrecked the database KDE uses for personal information management. Suddenly, KMail no longer worked, and I couldn't view KAddressBook.

A few moment's fiddling convinced me that I was not going to repair the problem quickly -- and I needed to work on something else immediately. My solution? I switched to GNOME for my desktop and installed Sylpheed for my mail, and in five minutes I was ready to start my other project. I had restored the functionality, if not my contact list.

This recovery, I realized would have been almost impossible on a proprietary operating system. Depending on the type of application, there might not have been a comparable redundancy. If there was, it would have to be paid for. Had I needed a better known app like WordPerfect or Photoshop, I might have been able to download it, but I might be able to try it first. In many cases, I would either have had to research shareware or free software (which, as a proprietary software user, I would probably know little about) or order something, which might take a day or two to order because software that was usually stocked.

Most important of all, I would have had no alternative desktop, and would have had to stop and make repairs, limped along with a crippled desktop, or fallen back on the command line (not a great matter for me, but a non-starter for many).

If I couldn't eventually repair the problem, I would have to reinstall the entire operating system, and then activate and register all my productivity software, in some cases having to plead the reason for my reinstall with a bureaucratic phone-answerer.

All going well, my system might have fully recovered after a couple of frantic and non-productive days. I know this would be the case, because a friend is just recovering from a virus on Windows, and her incoherent shouts of anger and sobs of frustration were heart-wrenching to hear and -- while justified -- seem unbearably prolonged to me.

The value of redundancy

This contrast shows the practical value of redundancy. Superficially, redundancy seems wasteful, and a common short-sighted view is that you can improve efficiency by getting rid of the redundancy. When all is well, that view may even have some validity.

However, in the long term, redundancy is a survival strategy, because damage to a part does not easily impair the whole -- even when the part is an important one. Alternatives to the damaged part exist, and can be swapped in with minimal delay.

By contrast, while a centralized monopoly may seem more efficient, damage to one of its parts can bring down the whole until repairs can be made. Chances are, too, that a centralized monopoly is less likely to be repairable without starting from scratch, since alternatives to a non-functioning part are less likely to exist, particularly if that part is essential to the system.

It's a difference straight out of General Systems Theory, and as applicable to a personal computer as to a cruise missile or a genome.

In creating free software, no one (not even Richard Stallman, so far as I can recall) deliberately set out to encourage redundancy. Yet the ethos that contributors should work on what interests them has created redundant systems all the same.

The concrete advantage

There is not much chance that free software projects are ever going to focus all their efforts on best of breed applications. The closest to this focus that has occurred is Mark Shuttleworth's efforts to encourage uniform release schedules for major projects -- and even that relatively modest suggestion could make no real headway.

However, contrary to what most people assume (including me, until recently), this lack of efficiency is not something we simply have to endure as inescapable evil. Instead, it is an advantage that free software advocates should be promoting in their efforts.

I suppose that the advantage is more or less covered by saying that free software is more robust, but that is a vague statement whose implications are easy to miss. Instead of talking about greater stability, we might argue much more convincingly if we say that a choice of software is not just an abstract good, but a practical one that means less recovery time and effort.

In much the same way, this redundancy is a natural and inevitable consequence of software freedom. Yet, while "software freedom" is an abstract concept that in this proprietary-dominated world means almost nothing to the average user, resistance to damage and quicker recovery time are concrete advantages that anyone who has ever crashed a computer (which is to say, almost everyone) can appreciate.

What's more, this is an argument that proprietary users may be particularly ready to hear. Many of them have seen the advantage of redundant apps in the web browser, first with Firefox and then with Chrome. In addition, during the last few years, many have learned to appreciate redundancy in the app stores for their phones. With such prominent examples, you should not need much persuasion to convince the average user that the same redundancy in any category of software and on any computer is a trait worth having.

All that is needed is a shift of emphasis -- and, suddenly, you have another reason for saying that free software is not just an ethical stand, but a very practical one as well.

Comments