@XavierP
Yeah, but here's my point; Out of your +18,000 posts, even if *one* percent were seen as helpful by others, you've out-paced someone like me, and therefore should show up as a "More Reputable" member than myself.
A percentage-based addition to the Reputation system doesn't do long-standing LQ members justice, IMNSHO*.

<not_sucking_up>
I was just using you as an example, because based on your posts in the past, I'll trust you to have the experience and know what you're talking about over any advice I would give (which is normally based off of a little experience and a few tests before I open my big mouth).
</not_sucking_up>

IMNSHO = In my not-so-humble opinion.

I'm not picking on XavierP either but, if *one* percent were seen as helpful, that means he didn't help 99% of people - he's wasting a lot of people's time. ;) Someone who bats a thousand is actually doing a better job, even if he helps fewer people. If that person helps you, you're one of the lucky few and should listen carefully. But under the current cumulative system, that person looks far less reputable. I agree that helping more people is a better thing and that an absolute percentage would perhaps be unfair to heavy-duty helpers but the pure cumulative system is just as unfair to light-duty helpers. It would be nice if the number of a person's posts or absolute count of helpful posts could weight in favor of them, but not so much as a purely cumulative system does. The system shouldn't be pure quantity, though it shouldn't be purely anti-quantity, either.

Right now it's almost pure quantity, though. If I post a thousand idiotic posts and accidentally get ten right, I'm going to have more rep than if I post nine out of nine good ones.

A huge number of my posts are standard mod responses or even hand crafted ones, I would expect not to be repped at all for them :)

Oh, I see, then I was using the right numbers. Ok, then. Still, I am against a percentage system, the numbers show the same trend.

I see that a percentage system says how helpful a person is, while the current system shows the total help.

I've added the LQ Helpful Quotient to the Statistics tab so we can further evaluate this. Note that this information was always available on that tab, we're simply doing the math and displaying it now. Additional feedback is welcome here.

--jeremy

Thinking about this further, there are actually two slightly different ways to generate the LQHQ, and I'm interested in which way gives a more accurate picture:

vs

--jeremy

I'm not sure. I suppose just observing some posters and their possible numbers under various calculation might give the best empirical answer. I think I lean towards the second one, but I'm not sure.

There are further calculation possibilities: you could factor in the aggregate positive/negative over the number of individual posts so that people could even exceed 100% if they had 5 positives and 1 negative over 3 rated posts, giving 133%. Or you could do the positive/negative over all posts, including posts that weren't rated at all, so that 5 positives and 1 negative over 20 posts (whether rated or not) would give 20%. This looks bad but maybe it could be converted into a letter grade graded on a curve or something. :)

I also like MTK358's post in #8 of this thread where it just shows positives, negatives, and neutrals as a spectrum. Then it wouldn't boil down to specific numbers but you'd just get a 'lotsa green, little red' impression on the more reputable posters. But I think something like that would have to calculate only the posts made since the system came on line.

Kinda bike-shedding it here. :)

When the helpfulness thing was first introduced I thought the idea behind it was to provide a mechanism to rate the content of post, not to rate the poster. Now it seems this stat is being presented as some sort of worthiness metric on our stats page. Mine is currently sitting at 94%.

The irony here is that in my case this 6 point drop was a result of taking part in the original discussion in LQ Suggestions about the helpfulness system when everyone was playing with the yes/no system to test it out. Now I never really cared about those handful of no votes sitting there on my stats page in the past, but as soon as you turn it into a percentage you're effectively branding me as "6% unhelpful" and I find that I do care about that.

This post is not about my specific situation, those negative votes will trend towards zero as things progress and I get additional positives ratings. I simply want to raise the point that unlike an open ended number like the rep system currently uses, as soon as you assign someone a percentage, you're not only saying what they are, but you're also implying what they're not. You need to be very careful about the psychological effects of this. No one like to be unappreciated and you could end up driving people away.

Well, mine is currently 89%, and a good part of it was due to arguments I had with others. Meh, whatever, I don't really care about this statistic.

Thanks for the feedback. The main intention of the Helpful system is indeed to help find quality posts. Some members were interested in seeing the data from another angle, and as mentioned this data was always available (albeit in a different format). This is something we're still very much evaluating, so additional comments/suggestions are welcome. Also keep in mind that you can completely disable the "Statistics" tab via your privacy settings.

--jeremy

I'm not that bothered with the stats being visible to others jeremy. It was just that while I never really cared about a few negative ratings when it was just a count, I was surprised to find that I did care when you put it in a percentage format. I have to admit that I care less today than I did on first seeing that stat but as I had a negative kneejerk reaction to it, I just thought it worth mentioning.

Going slightly tangential for a second: Have you ever considered extending the star rating system that exists for threads to individual posts? It seems to me that that would be a good way to achieve the goals you had for the helpfulness system and would be in a format the userbase should be very familiar and comfortable with.

We originally considered a 1-$x rating system as opposed to the current yes/no options, but went with the simpler and more widely deployed/accepted option.

--jeremy

fair enough. I'll shut up then. :)