Well I certainly agree with you that voting can provide well deserved positive feedback for a developer. Although my experience was that it was more likely to get a downvote on a patch than it was to get an upvote. But maybe thats because people didn't like my patches. :-)

Thinking about this a bit more I asked myself why I feel that commentless voting is appropriate for most of the site, but that it isn't useful for patches. There is a contradiction there that I can't quite resolve. Patches just seem to be of a nature where no feedback is better than trivalent response like a yes-no-ignore-vote. Its like some kind of Warnocks dilemma. If you downvote me is it because you dont like my code, or is it the functionality? Is it because I used an inefficient approach or is it because there is a flaw in my logic? With a normal node generally an author can imagine what aspect earned the downvote, with code the range of possibilities is so wide that it is just hard to tell.

Also the size of the sample pool makes a difference as well. With a normal post the number of people voting tends to cancel out spurious votes. When there are only ten voters or so anonymous -1 becomes a bigger deal.