I'd like to seriously weigh up voting vs endorsement points. Taking what both WA said and the feedback, I want to see how they achieve that goal: showing what people want. (I'll consider it a bonus if it also tells the developer/s if stuff should be worked on particularly soon.)
I'll make the following assumptions about voting (all of which I consider quite reasonable): the only place to upvote a suggestion is on its feedback page; the amount of things you can upvote is unlimited (you can upvote anything once if you'd really like to do that);
How people spend endorsement points varies: I know some people have waited for a single good thing to dump all/most of their 100 points into; I myself have my points spread wide and have 1-5 points in most things I support and maybe 40 in one big thing and 10 in another.
Even on facebook/imgur/stack exchange/tumblr and other medium where people are encouraged to "like" a lot of stuff, people only like stuff they actually like. I think it's reasonable to say that if someone likes something, they'll be happy with it being on the site! So, what people want and what they like is pretty much the same thing. (It just won't tell us how much they want it.)
If everyone votes on what they want and like, then we wind up with the most-liked (most-wanted) suggestions pushed to the top. This tells us what feature implementations would, in theory, satisfy the most people at once. That's pretty good!
Let's consider what the endorsement points do in that regard though. As in a recent circumstance, a suggestion with 500 points might have just, I don't know, five people who want it, or at least think they want it. Meanwhile, another suggestion sitting at, say, 328 might have, I don't know, thirty different people who want it, endorsing it with between 1-50 points each. I don't know those figures. A suggestion one person wants could sit at 100, while a suggestion ten people want sit at the same figure, or maybe just 50, depending on how many points they each feel they could put into it.
So endorsement points don't really tell us much at all about who wants what. They just tell us by how many points' worth people want a thing. Is that actually a useful metric to know? Since peoples' point spending varies a lot, we could have the feature the most people want sitting at a fairly low number of points, while a suggestion only three people want sits near the top of the roadmap, so I don't think that's actually all that useful.
Let's go back to that notion that votes don't tell us how much people want something. Do endorsement points do that? I don't think so, very reliably. I put 50 points into something I realy like --- I'd spend 100, but I don't want to drop all numeric indication of support of other things I like. Someone else puts 100 points into one other feedback because there's nothing else they're particularly interested in. Does that person want their thing twice as much as I want my thing? Not necessarily; they just had the points to spare. Let's assume we both want our separate things just as much: should their want count twice as hard as my want? I don't think so, we're both just two people who want a thing. Heck, I really want a few things I have only 5 points in. All this means is that because I want more than one thing, my want is counted for less numerically. You can only tell how much I want something by how much I'm spending on it compared to other stuff.
However... people have only 100 points to spend. To spend points on something new, I have to support other things a little less or drop support entirely. So the things I spend points on are probably the things I most want, but this still doesn't tell you how much or how immediately I want something -- except by how many points I spend on it compared to other stuff, maybe. You just have my top, say, 20 list.
I consider that a problem as well though. Our data on what people want moves over time. People might still want other features, they just don't have room to indicate it anymore in light of other stuff. Ok, they want that new stuff more, evidently, but we're losing valuable data. An old suggestion gets support from a couple of dozen people, it doesn't get implemented, points start to drain away from it because they don't see the point in leaving it there compared to more immediate stuff, — we lose valuable information about things people might still want.
The score at this point is:
- Voting tells us what people want, but not how much they want it.
- Endorsement points don't really tell us much about what people want nor necessarily how much they want it; they tell us a lot about an abstraction which sorta-relates to both of those, but doesn't strike me as all that useful on either metric. We still don't know how many people want a thing nor really how much they want it.
We still don't know how many people want a thing, nor really how much they want the thing. Neither system does that. I'd suggest, however, if we additionally give people a small number of (let's pick a fun metaphor) "high priority stickers", and tell them to stick them on the suggestions they feel they need most immediately (they get X new stickers per month and/or can move them around), then we get lots of data about what people want, and clear data about what people think is most immediately needed. We could have a thing 20 people like but 1 person considers high priority, and a thing 15 people like and 10 people consider high priority. That seems more useful to me than "328 points". (You'll probably almost always upvote the things you give high priority stickers to. High priority stickers are counted separate from votes; they don't add to the vote count in any way and are displayed publicly.)
In summary, I feel that a voting system with a way to indicate priority in addition to and separately from voting wins out over endorsement points in most ways: we get clear information, we get a lot of information, and it doesn't encourage information loss.