ongoing · OSI In the future though, I think pretty well all software innovation will be either open-source or inside a big server, because the business model for shipping closed-source software as a product is just too twisted and weird.

In fact I disagree pretty much completely, as there is a third group in a sort of mid-range that ends up closed source … and I don’t see working any other way.

Open source projects only succeed if they can attract a sufficient number of developers, and a sufficient amount of developer time. Developer time can come from fragments of “free” time - when the choice to participate comes from the individual developer. Developer time can be full time - when the choice to participate come from the institution for which the developer works. The effort level required of the project determines the amount of development time necessary for the project to be viable (i.e. OpenOffice requires far more dedicated developer time to be viable than Perl).

The defining characteristic that makes open source projects more or less viable is the ratio between the number of end users and the number of developers. When the ratio is large then institutions (large and small) are likely to find an interest in dedicating developer time to open source work. The ratio can be small when end users are developers and contributed of bits of free time are sufficient.

Open source projects fail when they cannot attract enough developer time. There are lots of failed or marginal open source projects.

In the mid-range there are software products with relatively modest numbers of users. Products that are successful in this range return quite a lot of value to their customers. It is not at all clear to me that products in this range could survive as open source.