I know that underneat it is +-. But psychologically it is very different.

I've been a slashdot user (under two different accounts) since like 2001. In my experience it isn't different at all. And my research into feedback and rating systems indicates that most people don't utilize the added flavor that hyper-nerds (I mean that endearingly, and I'll self apply it) insist is necessary. I personally know a half dozen people who will swear up and down that all Netflix needs to improve their recommendation engine is to allow users to specify which elements of a film they liked and why (nevermind how that would work for the moment). When you imagine a large scale recommendation engine does that make any sense to you? How many people, of the already small fraction who provide feedback, will offer detailed feedback? In practice it tends to be 1/10th to 1/100th of the folks providing the feedback. And even then we have some indication that it may be perfunctory. Netflix might gain a tiny bit of information about the preferences of heavy users and would likely incur a decent UX cost (and drop in overall ratings) as a result.

In the case of slashdot it's pretty simple. The system is designed to treat ranking as secondary to the informational content in the moderation. You aren't supposed to treat +1 informative as an upvote. That's reinforced by the low ceiling on comment ranking, the meta-moderation system and some of the culture there (back when there still was a culture). But the system just doesn't work and the costs are large. For most moderation/ranking systems the goal is partially to order content for the reader by some measure other than chronological. In large enough discussions, this may be impossible or uninformative as the reader has to sift between dozens or hundreds of +5 replies which may just have the benefit of timing. In audio terms, there is a tremendous amount of clipping and information is lost. As a result, you're back reading a flat list of (mostly) pruned comments. You can sort by other measures but almost nobody changes the defaults so I don't see the sense in analyzing the system by starting w/ the assumption that they do.

Another core difference which didn't end up mattering was the cost and scarcity of moderation points. The idea here was that if points were offered randomly and in finite quantities users would treat them as valuable and wouldn't bury comments or expend them on frivolous or jokey replies. I could just leave slashdot comments as an example proof of this failure but I won't. First, the random allotment. On ideal function of this was to prevent drive-by ratings just as we see on reddit. If someone sees my comment and says ""are you fucking serious" is rude" they may not have the power to downvote just that moment and they'll have to reply to my comment or ignore it. However as the community grows, the law of large numbers bites you in the ass. Even if a small number of individuals get the power to vote that ceases to work well once the post sees enough traffic. Start seeing the growth slashdot did from 2000 to ~2007 and you can imagine the results. The scarcity bit works the same way, with a twist. They added in expiration for the votes in an attempt to convince people to not hoard them and to provide some immediacy. That was a decent idea and a good attempt but it provided an interesting side effect. Once you knew that the points were going away, they started to burn a hole in your proverbial pocket. It was ok to spend them upvoting dumb things because they would evaporate in a while anyway. Like I said at the top of this paragraph, I could point to the variable quality of slashdot comments as proof this didn't work.

I have no idea how you'd apply stackoverflow's system to reddit. That would be odd indeed.

Yeah, but you can imagine the basic contours. It likely wouldn't work without a level of moderation that most reddit users (and mods) would balk at. But a system of tiered privileges based on karma might work fine. Especially if those were tailored to the demands of the site.