I have wondered for awhile whether the problem of post doc glut / suffering couldn't be alleviated by creating an alternative, more attractive avenue to the university science model. Universities charge massive overhead on grants which they use to support non-research related endeavors, place low ceilings on salaries, force investigators to do ancillary tasks like teaching, and don't promise tenure track positions.
There is no reason why someone couldn't create a for profit center which charged half the standard overhead on NIH grants making them very appealing, uncapped salaries, allowed investigators to own IP / arrange for favorable splits in a TLO model, and offered PI status to post docs if they could bring their own grants / industry money. Post docs would have a harder time staffing labs with grad students in this model, but they could get an affiliation with another university or simply hire students with BS's and pay them competitive rates.
Large research universities offer things that I think you would have a hard time replicating with this model. For instance, a research-grade fMRI facility costs millions of dollars to run and requires a large number of dedicated staff to maintain it. I can think of a lot of other resources like this -- primate facilities, certain kinds of DNA sequencing equipment, etc. Universities absorb this kind of cost with their endowment. I'm skeptical you could get a critical mass of scientists in a certain sub-speciality to be able to fund those things profitably, especially if they were doing basic science research (not pharmaceutical kinds of stuff that can make a profit on their own)
To echo Thriptic comment: you would be surprised to know how little top US research universities actually provide for their, well, research labs!
The way it works today, mutatis mutandis, is more like a "hacker space" than anything else: you get, if lucky, some shared shoebox as an office; a parked website domain maybe; and some rather vague and guarded promise of institutional support... But other than that you run your lab essentially like a startup: you are responsible for securing funding for equipment, salaries --yes, including your own!-- and
workspaces. And if you can't, you close down: as simple as that.
So the oft-cited 'advantage' of having a steady supply of quasi-indentured post-docs and grad students to horribly exploit is only half the story: if you just don't have the funding, you can't hire them. Note that in this (restricted) sense, throwing more money at the problem would indeed help towards the solution.
It's quite grueling and ridiculous when you think of it, especially in this era of billions and billions of dollars in endowment at your typical RU/VH...
I actually would not be surprised -- I did neuroscience research at two major research universities, and I handled the lab finances as part of my RA duties.
Our grants and the university-provided lab startup funds bought the smaller stuff -- EEGs, eye-trackers, etc -- which ran in the thousands of dollars. The university financed the building and some of the maintenance of the larger stuff like the fMRI facilities we used, and we paid hourly to use them.
At my university most infrastructure in core labs is actually purchased by individual labs and then placed in a core with the understanding that the core will pay for the service contracts and run the machines in exchange for sharing. The major costs are personel and service contracts. Moreover, many cores are able to be self sufficient by charging users for services.
A similar system could be established here, or investigators could contract work out to 3rd party companies or even other university core labs which frequently allow unaffiliated investigators access, albeit at a higher price.
Except MPIs are national labs and not for profit. The U.S. has Los Alamos, Brookhaven, Argonne, Lawrence Livermore, Lawrence Berkeley, Sandia, Pacific Northwest National Laboratories, Oakridge...
There is no reason why someone couldn't create a for profit center which charged half the standard overhead on NIH grants making them very appealing, uncapped salaries, allowed investigators to own IP / arrange for favorable splits in a TLO model, and offered PI status to post docs if they could bring their own grants / industry money. Post docs would have a harder time staffing labs with grad students in this model, but they could get an affiliation with another university or simply hire students with BS's and pay them competitive rates.