The Simons Foundation and Open Source Software

The open source community is in the wonderful unique position of having been mostly built during a time of instant and limitless free communication, and uses a set of tools and practices which are very accessible and not nearly so entrenched.

When you think about it, the speed of change and adoption in the whole constellation of modern software development tools – web browsers, search engines, email, usenet/mailing lists, internet chat, debuggers, profilers, package management, build tools, virtualization, graphical diff, version control and DVCS, blogs, wikis, ticket trackers, Q&A sites, etc. etc. (not even to mention various dynamic language features, coroutines, better garbage collectors, JIT compilers, network APIs, &c.) – has been remarkable, and software development today looks very different than it did 30 years ago.

The speed of iteration and experiment in development of tools is so fast because the people building the tools and the people using them are a similar and heavily overlapping group, and because building new tools can be done relatively quickly and cheaply by small teams, and distribution and marketing of open source are cheap.

Tools for, say, CAD/CAM, architectural design, structural engineering, 3d animation, video editing, chemical reaction simulation, etc. are much slower to evolve, because they take more professional expertise to work on, cost more up front to build, and have a set of users who mostly are not qualified to build the tools themselves and tool builders who don’t necessarily do professional-level work with it. Instead of “build what you need”, there are two disjoint groups whose mottos could be “buy what you need” or “build what you can sell”.

To take the example of a field I’m familiar with, image editing/printing software (along the lines of Photoshop), there are at least 5 disjoint groups involved in any innovation: artists, technical authors, user interface designers, software engineers, and academic image processing researchers. New features are typically developed as a somewhat abstract mathematical model by academic researchers whose incentive is to publish papers with elegant theory, implemented by software engineers whose goals are mostly based around performance and raw function, fit into the existing interface by designers who are trying to match long-established conventions, explained by manual/book authors who want to sell books (which usually means listing recipes instead of explaining difficult abstractions), and then used by artists who usually end up with no clue what the feature does. Along the way product managers, marketing departments, and others also get involved. None of those groups necessarily has the cross-cutting expertise to really understand the needs of everyone up and down the stack, so new features tend to be shallow/gimmicky, mistargeted, poorly designed, poorly explained, and ultimately used haphazardly or ignored.

But here are certainly room for improvement above Blog Comments as a collaboration UI. Even a google spreadsheet would have been better.

That's an odd way to describe people who effectively do their job of selling the product.

Over a million dollars to fund math software? Maybe, but not the demographic/popularity mix I would expect to see at KS. KS is for initial funding of a product that will be profitable at steady state. SMC is already launched, but needs more steady-state funding.

Prestige is the concentration of high quality research (as judged by peers) in one place.

I fail to see how making only some software available to US academics (which seems to be what you are suggesting), in any way enhances their prestige or capacity to influence science policy.

I wonder if NASA would get more funding from congress if we told them they were only allowed to use SAT-solvers in planning space missions.

But the notion that the Simons foundation is somehow trying to make software "selectively available", rather than just increase the availability of one particular piece of software they like, seems far-fetched to me.

See also: tragedy of the commons, and free riders. People like not paying for things, regardless of quality.

In fact the history of mathematical software shows precisely the opposite of what you claim. Companies go belly up and unless that mathematical software is open sourced, it dies completely. There are many examples of this. Sagemath in fact contains some code which came from a formerly commercial source (Maxima, formerly Macsyma), which was later open sourced.

You also miss that Magma is run by a group of academics at the University of Sydney, and there is real concern what will happen to it when the person running the project retires. Magma is not open source, and it is a very real risk that it will just die.

William Stein's understandable frustration is at the paucity of funding for a project as large as Sagemath, on which so many people are relying. He has written a series of blog posts over the last little while outlining his frustrations. It seems that his last post on the issue led to someone being "100% sure" the Simons foundation would write a cheque. William is demonstrating how very false that is. The assumption is that it should be easy to find a source of funding for such a large and important body of work. But counterintuitively, this has not proved to be the case.