Given the added burdens of...

- managing licenses across multiple devices

- limiting the people you can share code with to ones that are willing to deal with licenses

- budgeting for an ongoing subscription / renewing as needed

...I'd say that even $10/year is too much. Ecosystems that create collaboration barriers like these are no place for important work.

The support and documentation for MATLAB is really unparalleled though. Worth every penny when you want to focus your time on being productive, rather than banging your head against the language.

That sounds like a welcome change from the 2000s where we had to hound them for even obvious flaws.

Different worlds I suppose, but for me it doesn't count as "productive" unless my audience is successful. So having support for myself doesn't get me very far.

Without any toolboxes though, and you'd typically need a handful of those depending on what field you're working in and that's another $45 per toolbox. But also anyone just getting started may not know which toolboxes they need upfront which increases the uncertainty about the price.

I really wish they had a more reasonably priced all you can eat home license that included all toolboxes.

Agreed, it's sofa change for an institution. But I remember cost being a big obstacle when I first wanted to learn serious programming around 1990, so I feel like free programming languages are better.

The institutional cost is more AFAIK, but should still be pocket change -- one productive week pays for it. But the price is just high enough to trigger all sorts of weird management wastes. Especially if you envision wanting more than one installation, or are in an austerity-culture organization.

While you're dealing with the time and aggravation of budget approval and recurring license management, I've got my Python and Arduino toolchains installed on literally every computer that I touch: At home, my office, the labs, etc.

Sure, if you work alone, don't collaborate, and don't require any extra toolboxes.

I've seen various communities move from Matlab to Python, Octave, R, and Julia notebooks (that now support several languages). Ease of collaboration (including git repos and notebook interfaces), transparency of the platform (to tell what it's really doing), and ability to use computational resources.

I had a request from a grad student to run a simulation on a 40 node cluster. Said cluster cost $110k, including 40 nodes costing a bit over $2k each and some networking/storage. This would have reduced their embarrassingly parallel projects for their phd from 6-12 months to 5 to 9 days. Sadly the required matlab licenses would have cost more than $110k (the cost of teh hardware) because of the licenses and extra toolboxes. They cried. They had to delay their graduation and either reduce it's scope or rewrite it.