The Game Theory of Open Code [LWN.net]

Mikael Pawlo has written a paper that applies game theory to open and closed-source software models. "A company selling proprietary software to third parties will never open its code if the company has a competitor. It will never release its software under the GNU GPL. If you consider open code a benefit to society, you may want to propagate open code-legislation or otherwise try to stimulate new competition in the marketplace."

to post comments

The Game Theory of Open Code

Posted Jul 16, 2002 21:00 UTC (Tue) by DeletedUser2601 ((unknown), #2601) [Link] (2 responses)

The LWN summary could be misleading--this author does see scenarios where companies will open the source to proprietary products:

"A third company, Company C, may enter the market with a database solution based on open code, hence dissolving the Prisoner's Dilemma for Company A and B, and - if Company C is well received by the market - force them to open their code."

The Game Theory of Open Code

Posted Jul 17, 2002 4:56 UTC (Wed) by dwheeler (guest, #1216) [Link] (1 responses)

For an essentially opposite conclusion based on game theory, see Open Source as ESS by David Rysdam. It applies game theory concepts to software licenses and argues that the GPL fundamentally ``wins'' over other licenses. I mention this in my Open source software/ Free software (OSS/FS) references.

The Game Theory of Open Code

Posted Jul 17, 2002 9:36 UTC (Wed) by DeletedUser2603 ((unknown), #2603) [Link]

That is an interesting article, but using very different assumptions. In my example we have an oligopoly of propietary players. In the example you link to you have a functioning market (no monopolies) of proprietary and open code players. In that situation I agree that the open code solution will win. If you read my draft you will find this passage which argues for the same result as the article you link:

"A third company, Company C, may enter the market with a database solution based on open code, hence dissolving the Prisoner's Dilemma for Company A and B, and - if Company C is well received by the market - force them to open their code."

Don't waste your time

Posted Jul 16, 2002 21:07 UTC (Tue) by nas (subscriber, #17) [Link] (3 responses)

This guy doesn't have a clue. His grasp of game theory seems to
be limited to a few common examples of its application. He uses
Mozilla as an example of why companies will not be willing to release their
source code. Microsoft has not integrated all of the good features of
Mozilla into IE (as he suggests happens as soon as a company releases their
source). He says nothing about the GPL and derivative works. He doesn't
even try to quantify the costs and benifits to releasing source code. How
can you apply game theory when you don't know the payoff?

Re: Don't waste your time

Posted Jul 16, 2002 22:34 UTC (Tue) by DeletedUser2603 ((unknown), #2603) [Link] (2 responses)

As pointed out by Michael Sims on the FSL-discuss mailing list, Joel Spolsky argues convincingly that it's often in a company's interest to release open or free or Free software to sabotage a competitor and to make the things that complement what the company is selling widely available at a low cost: http://joelonsoftware.com/articles/StrategyLetterV.html

In my opinion, Microsoft did exactly that in the browser wars through price dumping and bundling schemes, but that is pretty far from Microsoft adopting any of the present open code models. I actually think that Bill Gates told the truth when testifying that [Microsoft] 'did not try to monopolize the browser market'. Sabotaging it for everyone else was good enough. But I think only very well-funded companies can apply strategies like that. Using a large amount of R&D on developing competing products just to kill of competitors seems far fetched for SMEs.

The situation I am interested in is another one, where an oligopoly is present with few players offering the same kind of off-the-shelf software product. The companies make a good revenue because of the lack of competition. Have you ever wondered why software is soo expensive? Why would any of these companies expose itself to the great business risk that opening the code represents?

It is easy for IBM to adopt open source and free software, because IBM is NOT an off-the-shelf software vendor or developer acting in an oligopoly. Microsoft, however, make a great deal of money selling calculus programs (Excel) in a limitied market. Excel could be vastly improved by the global developer community, should it be released under the GNU GPL. Still, Microsoft chooses to keep Excel proprietary software. Why is that?

If you are the only company opening your code in a limited market you will not gain from opening the code, at least not in a short-term perspective. Opening the code is a one-time decision, hence tit-for-tat decisions can not be made (I know the Nash equilibrium problem related to this simplification). The situation described is an oligopoly (actually it is a duopoly), very common in the proprietary off-the-shelf software business (i.e. QuarkXpress ./. PageMaker). Suboptimization will not lead to a release of the source code but a well-kept proprietary source code. Releasing the code may decrease development costs, speed up development, increase security and everything else that Raymond brilliantly observes, but revenues will diminish to a vanishing point (free is not gratis, I know, but still...) and may even kill the company. Remember - the companies are in the business of selling software, not hardware or services. Still, I think that in the long run Company A and B would benefit from releasing their code. But I can not see that happening without external, third-party intervention, because of the possible short-term consequences.

There is a twist, though, making your point possibly better than mine, proving me to be as you would put it - 'clueless' .-)

As observed by Marshall Van Alstyne, there is an "inter-network" effect wherein the demand economy of scale crosses distinct market types.

See:
http://papers.ssrn.com/paper.taf?abstract_id=249585

Best regards

Mikael Pawlo

Re: Don't waste your time

Posted Jul 17, 2002 1:00 UTC (Wed) by oconnorcjo (guest, #2605) [Link] (1 responses)

Actually it is when a company is on the brink of losing thier market to the "competitor", open sourcing some products looks attractive. Look at Mozilla and Open Office as prime examples. I actually expect to see more things like this in the future when a company starts thinking "What do I have to lose?". Mozilla was created in the hopes of stopping MS from ruling the WWW and Open Office was created because Sun can't sleep at night knowing that MS can leverage thier bussiness ap. market to get into places Sun is or wants to be.

Re: Don't waste your time

Posted Jul 17, 2002 9:39 UTC (Wed) by DeletedUser2603 ((unknown), #2603) [Link]

If you are losing the market share you will open the code - I could not agree more. This is the Netscape example. However, if you have stable market position in an oligopoly, you will not open the code without external factors changing the situation. That is pretty much the point I am trying to make .-)

The Game Theory of Open Code

Posted Jul 17, 2002 14:51 UTC (Wed) by DeletedUser2025 ((unknown), #2025) [Link]

I agree with the paper's conclusions, with one proviso. Its conclusions are limited to commercial software - software that must be sold by a company for profit. It does not look closely at software that is primarily a tool that complements companies' primary products or aids the companies' business processes.

When software is a company's primary product, it must be sold for profit, and opening the code removes much of the competitive advantage the company might have. In this case, it makes no business sense whatsoever to open the code. When the software is considered a tool rather than a product, however, then opening the code can really work.

Although the author glibly dismisses internal development, I believe that this is the primary area where open source will succeed. There are many companies now that are paying developers to work on open source projects - not to be able to sell the products, but to use the products as tools that complement their business processes or commercial products. Look closely at the developers involved in any major open source project, and you'll see evidence of this.

Even when software is developed internally, it often meets the needs of others as well. If it's open, then it can be developed and used by everyone who has similar needs. Look at the history of Samba, for instance. It was developed to meet an internal need, and then stagnated when that need disappeared. Since Andrew had opened the source, though, development resumed as others had the same needs. Now it's very widely used, both in businesses and home networks. Even with many competing software-as-products (PathWorks, LM/X, Lan Manager for Unix), Samba became popular.

The Samba example is not equivalent to the "Company C" scenario in the paper, however. Samba was not developed or released as a product. It was never expected to make a profit. Instead, it was developed as a tool, and the open code allowed everyone who had a need for the tool to improve it until it met their needs.

Ultimately, I have to wonder whether the idea of software as a product will die out in the end. When open collaboration can provide effective, low-cost, customizable software (like Linux, Apache, PostgreSQL, Perl, PHP, et al.), it's hard to imagine any way that software as a product can compete. If I am right, of course, this would mean an eventual restructuring of the whole software economy, with all the related financial and political turmoil that that would entail.

I think we can see the beginnings of this turmoil already, with the big push in many countries for legislation supporting proprietary software (like CBDTPA), or supporting open software. Ultimately, though, this battle will be won and lost in the trenches, by the overall utility of the software, not by legislation.