In the 90s campus to me was like a small city that was self-sufficient in a lot of ways. The school provided its own services in-house. A prof also told me he would teach us what industry is doing wrong so we can correct it -- that academia was ahead of industry. The school chose the best tools and languages for teaching, not following whatever industry was using.
These concepts seem to be getting lost. These are some universities who have lost the capability of administrating their own email service:
- mit.edu → mit-edu.mail.protection.outlook.com
- unm.edu → unm-edu.mail.protection.outlook.com
- ucsc.edu → aspmx.l.google.com
- ucsb.edu → aspmx.l.google.com
- cmu.edu → aspmx.l.google.com
- princeton.edu → princeton-edu.mail.protection.outlook.com
I have to say it’s a bit embarrassing that these schools have made themselves dependent on surveillance capitalists for something as simple as email. It’s an educational opportunity lost. Students should be maintaining servers.
These lazy schools have inadvertently introduced exclusivity. That is, if a student is unwilling to pawn themselves to privacy-abusing corps who help oil¹ companies find oil to dig for, they are excluded from the above schools if required to have the school’s email account.
Schools pay for MATlab licenses because that’s what’s used in industry. But how is that good for teaching? It’s closed-source, so students are blocked from looking at the code. It contradicts education both because the cost continuously eats away budget and also the protectionist non-disclosure. A school that leads rather than follows would use GNU Octave.
Have any universities rejected outsourcing, needless non-free software, and made independence part of the purpose?
- Google and Microsoft both use AI to help oil companies decide where to drill.
On a note on matlab, in addition to industry, there are certain fields in academia, eg neuroscience and many engineer fields, where matlab has been part of their culture for quite some time. My guess is you can make the case for some other proprietary softwares used in university. Changing culture in a field is not an easy thing; but fortunately people in science usually notice these issues and make a choice for themselves.
Plus, like you said, it’s used in industry, eg matlab in engineering and adobe in design. One argument one could make for university paying for proprietary software is that they get their students ready for the jobs in industry afterwards. So the teaching needs to be with these softwares. Of course, it would be preferable if they also offer education with the alternatives in the same course so that students can be more adaptable. But that can many times add more workload to already complicated concepts for the students to learn.
Plus, the world is larger and more complex than what it used to be. Whether we like it or not, offloading tasks to other entities, rather than completely doing everything by oneself, is usually the preferred solution, especially if the cost of implementation/adoption is high and those other entities have experience with such issues. The example is email, like the other commenter explained.
So, I think the universities see the needs for these proprietary softwares, either because the complexity is too high (eg email, per the other comment), or (some of) their faculty/students want it (eg matlab, adobe).
Thus, I don’t believe the answer is complete rejection. It should be that universities give people at least a choice in the matter, where possible; for instance, matlab and the alternative is python or julia. This is evident in HPC setup where they offer many packages rather than forcing students to settle with only proprietary stuff. And they should also advocate more open source and free alternatives; and usually university libraries do this.
When I studied compsci, my prof told me ½ of what I learn at the uni will be obsolete by the time I report to work. So his take was to give a strong dose of the kind of knowledge that does not expire: theory and concepts. We learned a language that does not even exist in the real world (PEP5), which was a blend of important constructs from several real assembly languages. He said if you learn PEP5, you will be best adapted to picking up any assembly language. If he were to teach a real assembly language the chances we would encounter it would be slim and we would be alienated by dissimilar other real langs.
The wise move is not to make students dependent on implementation specifics.
IIRC, the GNU Octave language is similar enough to MATlab that if someone cannot adapt something must have gone wrong with their instruction, which should not be centered around implementation particulars.
MATlab can only be justified in one niche case: simulink, which GNU Octave does not offer. A prof should have to have simulink as part of the course if they are going to justify spending dept money on MATlab.