Anarchist Development Produces Inherently Superior Software (Eventually?)

-- By MakalikaNaholowaa - 11 Jan 2010

Many people have a negative knee jerk reaction to the argument that anarchist software development produces superior goods. The opposite position, that proprietary software can be superior to its free counterparts, is defended by the same argument raised in our class discussion; closed software is perceived as more feature rich. I have always wondered if people with doubts about open source superiority would agree with the anarchist development principle if it were amended to include two words: over time.

Open Software Gets Fixed and Improved Better Because It Is Fixed and Improved By More People

A Car Analogy

Free and open software is inherently superior over closed for the same reason that cars would be better if every driver was also an engineer and a mechanic. Drivers would identify flaws and unacceptable product limitations, and then modify the vehicles to resolve these shortcomings. More efficient, safe, and feature rich cars would result. And not only would the cars be improved, they would be superior to cars only improved upon by the original manufacturer's staff, because GM's 244 thousand employees could not out perform America's 199 million drivers.

The same idea applies to software, except it is much easier for a user to learn to modify software than it is to become a mechanical engineer, and unlike with cars, the cost to modify software is virtually zero.

Although Superiority Is Virtually Assured, Timeframes Are Undefined

Though open source development will eventually lead to more stable and reliable software, the timeframe within which specific features will be developed by the large but disaggregated open source community is undefined and in many cases longer than in the closed source space.

For example, image editors are ubiquitous in the open source world, and there are few things you can do with the closed Adobe Photoshop program that can’t be done with the GIMP, an open source editor. But 14 years after the GIMP’s first public release, a small feature gap still exists, and it is significant enough that Photoshop continues to be the industry standard for image manipulation.

Admittedly, some of the feature differences between the GIMP and Photoshop exist for reasons other than simple lack of development by the open source world (i.e. proprietary formats preventing RAW file format support in the GIMP). But some features could be addressed by development, and we just haven’t gotten there yet. For example, the GIMP does not natively support the CMYK color model, an important and long time requested feature that is needed for precise printing.

GIMP supports CMYK color selection internally. It does not natively perform CMYK separation, so you can't do prepress work on images that are going to be printed with traditional 4-color processes. One of the reasons this is not a high enough priority to go from a plugin to the main line code of the GIMP is that so much is now printed using inkjet printers. Those working constantly with color offset printing don't use the GIMP, and those who need it occasionally have the facility to do so.

This functionality will eventually be available (“experimental” plug ins already exist), but print shops aren’t holding their breath while they wait. Why? Because this example illustrates that open source development eventually produces the tools people need; over time people will create the tools they need. But there is less pressure for the community to create tools within a specified time-frame. To the extent that superiority is determined by comparing how fast tools and their feature sets are available, many times proprietary software wins the race by a mile. When this occurs, open source tools struggle to displace their closed source competitors.

There's no guarantee that proprietary software will be faster to include particular features than a free competitor. IE will never catch up with Firefox's feature set given the ease of making add-ons for Firefox, for example. Simplicity and compactness can be a feature too, and only a free version of something can be built for speed and light weight very often. Consider also the programming language development activity, which is hardcore programming at its finest. All the real innovation in the last generation has occurred on the free side since Java: Perl, Python, Ruby. The most exciting new thing going on right now is probably Mythryl, another free project. Who is doing the finest proprietary language development? MS puts Ruby on .NET?

MySQL’s Future Due to the Oracle-Sun Deal Further Questions Anarchist Development’s Ability to Keep Pace with Closed Source Competitors

The European Commission is investigating Oracle’s acquisition of Sun Microsystems. Sun owns the copyright and trademark rights to the MySQL database server, one of the leading tools in its product category and major competitor to the Oracle database server. If Oracle is allowed to purchase Sun then the EC’s concern is that Oracle will discontinue support and development of MySQL, which could decrease or eliminate the significant competition presented by MySQL against Oracle’s flagship product.

Michael “Monty” Widenius, the creator of MySQL, is asking the public to object to the deal. Monty fears that Oracle will indeed stop supporting MySQL, thereby “killing” it. Other experts have provided strong reasons why Oracle is unlikely to do this. But, if Monty is right, then the resulting impact on MySQL further raises the question of whether anarchist development can keep pace with closed software creators as needed to maintain market strength.

MySQL is currently available under GPL2 by Sun, meaning that all of the software’s existing code is available for use, modification, and redistribution. In a worst-case scenario of sorts, after acquiring Sun Oracle could discontinue supporting MySQL and stop releasing it to others under the GPL. But today’s holders of the tool could create a “fork” of the database server, continue to develop and support it using the code foundation available today, and distribute it under a new name. To some, this seems like a mild worst-case scenario that makes Oracle look like a small threat to MySQL.

However, there is a real possibility that because of MySQL’s size and complexity, even with the ability to fork, MySQL will slowly fall into disuse without significant resources and leadership by a managing organization. As Monty discusses at length on his blog, the fact that the source code is available and the open source community is licensed to use it probably won’t be enough to keep MySQL competitive against its closed competitors. At base, the problem seems to be time.

Without quick and large investment by an organization willing to lead the creation and support of a fork with full time development, the task will fall to disaggregated developers investing varied time to the project. During this time, fork-!MySQL is likely to struggle, losing user trust and reliance, while its proprietary counterparts move forward with new features. Over time individual forks could continue to improve and add features, but under fragmented development conditions, a fork-!MySQL is unlikely to maintain parity with its closed source competitors.

MySQL has no closed source competitors. It's the world's most installed database, bar none. It's the only tool of its kind that is known to every technology-literate fifteen-year-old on earth. Its unparalleled mind-share makes it an enormously important asset for anyone, even Oracle. Moreover, its complementary use to Oracle 11 in every large-scale data-center facing the web, whether in telecomms or airline or financial services or anywhere else, means that Oracle has immense service revenues to look forward to as the best place to get not only your big warehouse software serviced, but all those clusters of customer-facing MySQL servers that pull from the big warehouse in order to put your data where the customer can see it. That's worth tens of millions a year to Oracle, which Oracle will largely invest in killing the only database product that competes with it on the high end that can also scale down to where mom and pops can possibly use it: Microsoft SQL Server. Acquiring MySQL allows them to sharpen it into a dagger to plunge into the heart of SQL Server, while getting the work paid for by their own current customers. They don't want to kill MySQL, they want to make it shine.

Monty, as it happens, is—in my view—wrong. We've talked about it a great deal, both in public and in private, and I'm afraid that's the conclusion I've come to. His views depend on some intermediate conclusions that I don't think the data support. He is worried about Oracle's future behavior for quite comprehensible and laudable reasons, but I think he's seeing it wrong. Also believing he has it wrong are his former CEO, Mårten Mickos, and Brian Aker, one of the earliest and most active developers of MySQL, now leading the project to adapt MySQL to the cloud era, called Drizzle. Brian is presently at Sun and will not be joining Oracle, and he knows that Drizzle can survive as a pure GPL descendant of the MySQL codebase. I'm not carrying an Oracle brief, nor are Mårten or Brian. Monty's analysis is clear to us all, and we understand we he says what he says, but we don't think he's right. As you know, this has been much discussed in public and in pseudo-public (at the European Commission) in recent weeks, and the general consensus has come to be with us.

Conclusion

Anarchist software production will produce superior goods. But sometimes, the trade off for superior tools means waiting longer for specific features. When that happens, open source tools have difficulty competing with closed source options.

I'm not sure you've shown that this is the trade off, or that the competition between free software and proprietary software is feature-driven. That's not obvious, given that price is often important and so is freedom.

I agree with your analysis of the convergence to (1) technical reliability, and (2) fitness for use by all those with a significant interest in adapting the program or having it adapted to their needs. Quibbling is of course possible over whether those constitute superiority.