The standard rate for overhead billed to a grant is 30% and this has increased over time as has the amount of university administration and their salaries , so almost certainly we're not dealing with increased physical plant costs(typically, specific research equipment is billed to the grant, overhead covers things like building maintenance and admin costs).

So yes, (some) science is expensive, but universities also profit from it.

This really isn't true. Math and computer science research need very little in terms of materials cost. The physical sciences are indeed more of a problem, but I have difficulty believing that everything needs big money to research. Surely a large communal lab would facilitate a lot of smaller scale work?

Those fields aren't the issue, and not what he is discussing in the article. The two fields he mentions, bio-chem (the DNA sequencing) and quantum physics (higgs), both require large amounts of funding to progress. Think the large hadrom collider was cheap? Think a DNA sequencing machine is cheap?

Math is only limited by computer power, and our own personal mental computing power. Computer science, similarly. Those two fields are different from the rest of science as we know it today.

It's true that certain aspects of science are very expensive. We're generating a lot of data these days through DNA sequencing, particle collision, and various other expensive scientific measurement techniques.

But there's a lot of work to be done in making sense of that data. It doesn't cost anything to play with data that's already been collected.

The biggest breakthroughs come from people thinking differently.

Not sure what you're trying to say, do you believe collected data is currently shunted to a bin while the next sequencing run is started or something?

No. I'm just saying that we're collecting data faster than we can make sense of it. New discoveries don't necessarily require new data.

Obviously experimental physics can be extremely expensive, but coffee-and-chalk theoretical physics positions like (afaik) Higgs' have roughly the same costs as most math or humanities posts. And apparently theoretical physics has huge problems of PhD oversupply.

Does this disprove your implicit point though? The problem is any position which doesn't cost a lot is attracting way too many people for the amount of possible work.

Meanwhile, the expensive parts - i.e. all the applied physics needed to prove or disprove theoretical physics, is not receiving appropriate funding.

That's a really good point. Many meaningful projects will eventually cost a lot of money.

But I like how Brenner provides an environment for scientists to basically do what they want. Some of them eventually get projects funded, but it's important to have that first stage in which people have the freedom to explore.

A basic income can provide that freedom for everyone.

The problem with "the problem" is that there is no uniquely specified problem. The article is concerned with the use of key performance indicators by administrators to allocate resources to scientific projects in many fields. Math and computer science are also affected by this, contrary to what you say.