One of the effective solutions is to construct a risk table, in which you list out:
- each option - the costs to perform the option - the likelihood of success of the option - the value of success via the option - the penalty of failure via the option - the cost to ameliorate failure of the option
And you never assign a 0 or infinite value to any of those factors; a range of .05 to .99 is usually appropriate.
For binary choices, this can sometimes be short-circuited by flipping a coin and asking yourself if this feels worse than the alternative.
You wave away my association of "reading _too_ widely" and "analysis paralysis" by sharing your preferred definition of the latter ... but that isn't really an argument against the association of the two.
I didn't say that analysis paralysis causes reading too widely. I said that they are the same, so if you want to take your definition of analysis paralysis as a given then you could say analysis paralysis can cause reading too widely.
Anyway, all I wanted to communicate with my comment is that when you say that you can do "too little of X" but not "too much of X" then you are biased on X, there is always someone who takes X too far and you may not see it because you've associated those people with something totally different.
Personally I don't believe in "wasted time" or other subjective notions (in the abstract) so for me this whole discussion is lacking context (you can waste time if you have a context; some loss function, otherwise no).