A ranking by popularity might help finding useful postings.

I beg to question the premise :)

IMHO, the popularity of a node as measured by the voting system is not necessarily related to its usefulness for solving the problem you set out to find nodes for.

Even if all monks voted with good principles in mind, there would still be the exposure factor having a considerable influence on the final rep of a node. This is not only related to whether a node is frontpaged or not, but even more to when a reply is being posted. I've seen quite a few excellent replies being posted one or two days after the thread had been started, and in most cases they only got very few votes (below 5, if any at all). Had they been posted immediately as the first reply, I'm pretty sure they would have 'scored' a lot better. In any case, those nodes would definitely fall through the cracks when you rank by popularity.

Another factor influencing rep is how easily a node can be understood. Nodes that require too much thinking or reading get less votes on average (of course there are exceptions). More complex and lengthy explications thus suffer from this penalty, irrespective of how well and to-the-point they're written — some issues just require more words or code. On the other hand, it's exactly those nodes that might help solve your problem, rather than the umpteenth repetition of trivialities you can also find in the regular docs.

As far as I could observe in the time I'm here, both of these factors have a high influence on the rep of a node, but are virtually orthogonal to potential usefulness.

Note that I'm not complaining about the voting system (for heaven's sake, leave it as it is), I'm just saying that rep might be a bad measure of usefulness or 'quality' of a node.