Just a mini thought here.
In economics, there are a number of theories on how entities invest. One of those is has to do with the relationship between the amount of investment and interest rates.
Essentially, theorems such as the marginal rate of efficiency or the marginal rate of investment state that firms continue to invest until the marginal rate of return is no different than the interest rate. In short, when making a decision on where to invest money, money will always be invested in the investment that produces the highest rate of return, until there are no options that differ than the prevailing interest rate.
By analogy, when governments are deciding how allocate scare research dollars, similar economic decisions should be at play.
In reality, these decisions are nothing close to rational, in the economic sense.
For one, we cannot quantify the rate of return on investing research dollars. Comparing two proposals for funding, reviewers likely can estimate the rate of return. The equation is something like: # of publications x impact factor of likely journals**.
**This formula is likely discounted a bit for age. Old researchers are more likely to publish, but have a shorter lifetime return on investment. Funding young researchers before tenure makes it more likely they will be publishing in 10 years...
At a broader scale, this type of analysis is impossible or irrelevant. Funding agencies need to decide whether to invest in one discipline vs. another. This equation is useless for deciding whether to invest in physics vs. biology, for example.**
**If it was used, it would just favor the more prolific discipline. Publishing papers is not the likely goal for funding agencies, per se. Papers aren't bitcoins.
The number of likely citations is also not a good metric, because it's circular. The number of times papers in a discipline are cited on average are determined by the number of papers published in that discipline (and the average number of citations in a typical paper).
If we cannot use expected rates of returns on publications, then how do we assess worth?
This is the Achilles' heel of rational investing in science. We cannot find a common currency to evaluate the relative worth of research.
Without this, funding becomes irrational.
Right now, changes in funding levels occur at the margins. Politicians and directors do not regenerate funding levels each fiscal cycle. Instead, they make a case that funding for one area should be increased or decreased relative to what it is currently. There are no equations that are employed to determine relative funding levels or relative changes in funding levels.
Until we have a concrete way to assess the value of research, either in terms of dollars, or social equity, or longevity, funding is likely to be irrational.
Stepping back, not only the relative amounts of funding, but absolute amounts of funding for research need development. In the US, NIH budgets are 4x higher than NSF. Why the relative difference is one question. Another, is why are there combined budgets almost $40 billion? Should they be $20B or $80B?
If the decisions are fundamentally irrational (economically), then we either need to make them economically rational, or commit to irrational (economically) arguments.