[roel_v]

Because of course you would write some software that CERN, with 100's of engineers and scientists, finds not cost-effective to build its own version of, on a sunday afternoon...

The naivete and total absence of awareness of how things work in the Real World in this thread make me shake my head. 'Hey all, someone should write an open source version!' - sure, and in the mean while, those same people gush over some guy raising 300k for "the next generation code editor" who then barely manages to build something that works at all, let alone is an actual improvement.

Software is hard and expensive, especially software that has actual domain knowledge embedded, unlike the 1000th 'micro js framework for single page apps' and fads like that. All the hate on people and companies who make a living selling software (gasp!) is... well, there is no other word for it - plain dumb.

(yes, I make (part of) my living from selling software)

[JupiterMoon]

Generally this type of software relies upon inertia and vendor lock in rather than actual technical advantage.

[roel_v]

Yeah - no. Do you have experience with high end scientific software¿ Because I have, and that sort of software is hyper niche and requires highly advanced domain knowledge as well as expert software engineering, and is usually the culmination of several man-decades of research and development. If it was truely the easy to replicate cash cow some seem to be assuming, there would be competition all over the place; or a lab would get some intern to replace it in 6 months. But they don't because it's impossible and they realize that Adam Smith's point about specialization holds for software, too.

[tripzilch]

Right. The university paid a high price for MATLAB licenses (fine go ahead and claim that's not "high end" enough for your example), so they're going to teach their students MATLAB, even when the free and open source combo of IPython+Numpy+Scipy+Matplotlib is at least equivalent, in many aspects better, AND uses a saner scripting language (Python) than MATLAB's with all its crazy warts and lack of modern (15 years ago) programming paradigms.

> If it was truely the easy to replicate cash cow some seem to be assuming, there would be competition all over the place; or a lab would get some intern to replace it in 6 months.

Okay, now I have to ask. Do you have any experience with interns at a university trying to write even, let us assume, a relatively simple and straightforward piece of software?

Don't get me wrong, I partly agree with you, nobody's going to rewrite the high-end scientific software in a weekend.

But you seem to be saying that's because it's too hard. What I'm saying is that the people with the domain knowledge (physics students, interns, PHDs, whatnot, even the computational science guys) are pretty much uniformly shit at writing software. It's a big reason why there's not a lot of code sharing going on in the scientific community: Shame. And that if it was published nobody would want to read it, and there's a rather big chance it would reveal fundamental research errors.

The real reason this "high end" scientific software is guarded so heavily is because the fact that they have other users than themselves, they know they are in possession of a unicorn.

Then there's the part where all the old professors have their (unreadable) scripts written for that ancient unicorn, so even if something better comes along, nothing's going to change much. And even if the students translate (read: completely rewrite and fix some longstanding bugs) those scripts to the modern software, the professors won't have anything to do with it because they don't want to learn the newer thing since they spent so much effort getting to know the arcane old thing (though they will claim they don't trust the new software yet).

Dinosaurs. People will wax nostalgic over having worked at (gasp) actual UNIX terminals in their university years. Except it was the year 1999 and you had to FTP to the one machine with a floppy drive to take your stuff home. Sorry I digress, but yaaaaay science.

[JupiterMoon]

Yes I have experience. That is where my comment came from actually.