Going back to the notion of making things slightly (subtly?) more difficult --
What if the "thank" function were harder to use? More information could be attached to it, for example:
"You wish to Thank this post. What did you like about it?"
Funny
Polite
Insightful
Technical
Well explained
etc.
Maybe even adding traditional 1-5 scores or something.
Those data wouldn't necessarily even need to be used anywhere, the point being just to give the thanker something to think about before clicking the button. Obviously there's nothing forcing that to be the case every time -- but in the average case, people do tend to do what the thing says.
And so, it's no longer just a throwaway action, it at the very least demands enough attention to click through these things, and it can be that the answer is only accepted once enough options have been filled out. Even if one clicks through mindlessly, that can be correlated against e.g. per-user stats (low-value thankers get in a habit of ticking all 5's or what have you?), per-post or thread stats (is a post/thread just consistently bad -- in constructive ways? or is it being brigaded? Or, conversely, is a post/thread consistently good and perhaps worthy of automatic labeling as such?), and this provides a statistical mechanism to infer experts: those that are frequently reviewed, positively, in technical subjects (it could further be clustered by topic), by users that frequently give quality (high entropy?) reviews. Thus, the scientific method of consensus, but formalized algorithmically, independent of users' egos (hopefully).
Obvious downside: someone has to add those columns to the database, and write and run the evaluation. (Which might run incrementally per post, or be totalized periodically, and whatever cache awareness / synchronicity has to go with that, etc.) So, unless someone's already written a plugin to this effect, doubt it's going to happen here. And the value functions aren't the most obvious, and would require testing many variations to arrive at the most useful stats.
And then, Idunno, if it ends up written as an SMF plugin kind of thing, could that be sold in turn as a package to other customers? Maybe there's a value proposition in there. Maybe the value isn't anywhere close to what is needed to justify that amount of work (some months of development at least, I would guess?).
As for honest indicators, real financial barriers tend to be representative. Something of the neoliberal dream, put your money where your mouth is, quite literally. This might resemble for example Reddit's awards system. Which is generally not very highly rated as I understand it; but clearly, not so poorly that people don't use it. And, again, whatever financial liabilities the site would have to undertake to manage that. Maybe there are ways that e.g. Paypal can streamline such interactions, maybe there's just too much work to do (including international transactions!) that it's only ever feasible for the biggest players (i.e. Reddit themselves).
Tim