Don’t assume that most concepts have definitions. Look and see. How many of your concepts can you define? If you can name more than a few, you’re not thinking hard enough.
So when I say that the seminar is about Institutional Irrationality, don’t assume that “institution” or “rationality” can be precisely defined. We can work with those concepts, anyway. Still, the attempt to define them is worth the effort, because it increases the chances that we will be talking about the same thing when we use those terms.
Is 'rational' really the best term? It doesn't matter what the definition is if one is not using the apt term to describe the observations one finds troubling or mysterious.
My impression is that 'irrational' is something of a more neutral and innocent-seeming euphemism instead of more negatively loaded and culpable terms like 'corrupted', 'politicized', 'incompetent', and so forth.
So, one may be irrational by sincerely pursuing certain goals, but doing so in a way without any good logical connection to evidence about causes and effects or making other kinds of cognitive errors. "Sunk cost fallacy" is a good example of people behaving irrationally, indeed, it is sometimes called "irrational escalation of commitment".
On the other hand, if an institution simply purports and publicly claims to have certain ends and purposes which were traditionally genuinely pursued in a rational manner, but those goals have been secretly displaced by the rational pursuit of alternative agendas and ulterior motives, then the institution is not really behaving 'irrationally' except perhaps to the most naive observer who takes the stated mission seriously without any healthy skepticism.
Instead, you have the institution behaving rationally, but *dishonestly*, in terms of admitting its true commitments and what the people who are running it are really trying to do.
An antecedent point is that agreeing on definitions increases the odds of agreement only if the parties are committed to neutral ideals like truth, fairness, etc. across a range of applications for the subject to which the definition is being applied. But that condition hardly ever holds when it comes to contentious issues, because the issues are contentious due to preexisting normative (usually community- / identity-based) commitments. So the question becomes how to intermediate between conflicting groups that have a stake in changing or ignoring definitions when convenient. This is a question more of power and sociology and law than of language as such.