58

I wonder what is "People helped" (PH) stat supposed to signify. From what I read in Profile Page Makeover, Part 3: the Prototype returns, it stands for cumulative number of views on some (all) of my posts. I would like to know why we consider views as such a significant thing to put it on the very top of the profile page.

It has been discussed before that various things affect the votes in a bad manner: Hot questions, Meta discussions, Tweeted posts, ... This is of course based on the fact that these questions are viewed more. If it affects the rep to the extent that we request something to be done to prevent this from happening, why is PH introduced which is even more sensitive to this phenomenon?

As well, we've already got one measure that is ridiculous: the reputation. And what does it signify if I have 1.5k rep and 15k PH, while someone else has 1.5k rep 9k PH and yet another person has 1.5k rep and 33k PH? In my opinion, nothing.

In conclusion: I don't directly propose PH to be removed from the new profile page. I'm just eager to see any true use of it.

Edit: I made a very wrong but somehow working PH calculator on SEDE.

Improve this question
22
  • 4
    The idea is to better highlight that the real difference you're making often comes from the many people who find your helpful posts down the line. We'll definitely post the methodology - it's loosely designed to count views of your helpful posts, but it's a bit more complex (and we're still tweaking it). I'm not at a real keyboard today, but we'll get more color up in the next day or two.
    – Jaydles
    Nov 27, 2014 at 19:16
  • 7
    @Jaydles Thanks. I just hope you've well considered that fact that it's a second ridiculous number that people're going to hunt for in any available way, and it drives everybody from the real idea, which is sharing knowledge.
    – yo'
    Nov 27, 2014 at 19:32
  • 4
    No problem! We agree 100% that it's all about sharing knowledge. The idea is just to help you get s rough sense of how far your efforts go to get that knowledge to others.
    – Jaydles
    Nov 27, 2014 at 19:34
  • 6
    ... and I doubt that I really helped all the 15k people you count in. For me (and for many others who don't care about all these stats) it's just another meaningless number :-/ I see a lot that caring about rep (and other stats) doesn't bring anything good at all.
    – yo'
    Nov 27, 2014 at 19:37
  • 1
    @Jaydles did you consider showing PH-index in logarithmic scale? (I am not proposing that it's necessarily better than current linear, merely wonder if this option was considered)
    – gnat -on strike-
    Nov 27, 2014 at 21:29
  • 4
    Related: meta.stackexchange.com/questions/165179/…
    – Shog9
    Nov 27, 2014 at 22:54
  • 3
    I agree with Jaydles that it is encouraging to get a sense of how far our efforts have gone to help people. I helped 58K people in 11 months on SE, seeing that encourages me to write more posts. Earlier, I was janitoring more and contributing less. Now, I might actually start writing answers again.
    – Infinite Recursion
    Nov 28, 2014 at 6:19
  • 21
    @InfiniteRecursion I'm sorry to burst your bubble, but you haven't helped 58k people. You've gotten that many views. No one knows how many you've helped. Having my profile blatantly lie in order to boost my own ego is rather off-putting.
    – user206562
    Nov 28, 2014 at 13:18
  • 9
    @Jaydles So what's the point of reputation if not to track this very same thing? If you really think some agglomeration of page views indicates helpfulness better than upvotes, shouldn't you just change the reputation system rather than introduce a parallel one?
    – user206562
    Nov 28, 2014 at 13:21
  • 4
    @ChrisWhite, there are a couple of key differences. This number is designed to give you some perspective on how many times your content was seen by someone (presumably looking for help), but rep conveys privileges, and the latter requires that the community empower you. A simple example would be an (on-topic) trivially easy answer. It may help thousands of new learners who google it, which is a wondrful thing for the poster to know. But it might only get (or deserve) a few upvotes, and the tons of views shouldn't earn you the power to delete posts.
    – Jaydles
    Nov 28, 2014 at 14:16
  • 1
    Thanks @ChrisWhite for bursting my bubble.
    – Infinite Recursion
    Nov 28, 2014 at 14:29
  • 1
    @Oded: I appreciate this is discussed with a dedicated question now. But did you have to delete my previous answer over this? meta.stackexchange.com/questions/244238/…
    – Erwin Brandstetter
    Nov 28, 2014 at 15:35
  • 11
    @Jaydles I appreciate what you're trying to do, but I don't know if it'll actually do that. Also, consider the reverse -- "gosh, I've been on this site for a whole month and I've only helped 7 people? May as well give up then". I don't want to discourage users when they're just starting to contribute. Finally, if you're going to have this, you should probably give extra weight to visits from post-specific URLs (e.g. people who followed a link specifically to your answer).
    – Monica Cellio
    Nov 28, 2014 at 19:31
  • 1
    if I read correctly, closed questions and answers to these count. I would like to better understand reasoning for that. Some related stats wouldn't hurt either, like how much of average PH would come from closed questions
    – gnat -on strike-
    Nov 28, 2014 at 21:08
  • 1
    I don't think users on SE are that acidic or alkaline. :D
    – ʇolɐǝz ǝɥʇ qoq
    Apr 30, 2015 at 2:53

6 Answers 6

Reset to default
64

The method of computation of PH index equates helpfulness with traffic brought to the site. It is about SE being helped with traffic growth, not about people. It would make sense to call the PH index

  • Total viewership/readership, or
  • Traffic generated.

But "helped" is an outright lie. Internet search involves looking at things that don't turn out to be helpful. On Math.SE (where I have access to 10K tools), 41.5% of anonymous feedback is negative: those visitors explicitly said the content was not helpful, but are counted as "helped" anyway.

Lessons taught by PH index:

  1. Editing poor questions into shape (which prevents their closure and improves their discoverability) is not helping people.

  2. Searching for an existing canonical answer and marking the question as a duplicate is not helping people.

  3. If instead of doing 2), you post a half-baked answer of your own to a much-duplicated question, that is helping.

  4. Posting a single funny screenshot helps 136000+ people. Posting a funny pic copied from the Internet helps 350000+ people.

  5. Spending hours on review of first posts, edits and closevotes is not helpful, however.

On sites other than SO, traffic correlates with how bikeshedy the question is. When the "people helped" are bored SO users finding a diversion from their task in the SO sidebar, the "helpfulness" rewards clickbait titles and wtf question topics.

Or perhaps I am wrong and it is indeed very helpful to know how to draw a pair of buttocks?

Added after Jaydles commented:

Our whole system is based on your effort mattering...

The PH index is not based on that.

  • Peter Mortensen's titanic effort (editing 30000+ Stack Overflow posts, and thousands on other sites) counts for 0 people helped.
  • Asking "how to convert String to int in Java?" counts for 1600000+ people helped, even though the asker did nothing special at all. The same question could be asked by anyone else (in fact it was, many times).
Improve this answer
13
  • 7
    PH inventors probably think The Trouble With Popularity does not exist...
    – gnat -on strike-
    Nov 28, 2014 at 9:29
  • 5
    ...also, "smart" users will likely exploit hot network questions to pump up their PH index by dropping their answers into these. This guarantees few hundreds extra views... not more than that as system often fails to identify really entertaining questions but it's okay, brainless formula feeds new questions into sidebar every day, making it a stable "source of income"
    – gnat -on strike-
    Nov 28, 2014 at 10:12
  • 1
    Even if you take a pretty cynical view of our motives, remember that we get the pageviews whether the poster knows how many people found what they shared or not. Our whole system is based on your effort mattering - the time you put in here actually makes a difference to many people searching for answers, and we wanted people to have a better sense of how far their generousity goes. (I need to write up what inspired this count - there's a personal story that may help give us a smidge more credibility, but I'm sneaking these in from family holiday time.) :)
    – Jaydles
    Nov 28, 2014 at 14:10
  • 6
    @Jaydles And the problems come from the fact that you want to measure the time investment, but you only aim to measure (second time, rep being the first) some number that has a little correlation with the time investment.
    – yo'
    Nov 28, 2014 at 15:12
  • 1
    It's not time investment we're going for here at all - it's the impact your investment had.
    – Jaydles
    Nov 28, 2014 at 15:19
  • @Jaydles Citing you 3 comments above (emphasis mine): "the time you put in here actually makes a difference to many people searching for answers". I'm not trying to play a words-game, but I really lost you.
    – yo'
    Nov 28, 2014 at 15:50
  • 1
    @tohecz just shift the emphasis to makes a difference!
    – Jaydles
    Nov 28, 2014 at 15:57
  • 18
    @Jaydles But I still don't see how this number makes a difference. I don't argue that we want metrics which measure that my contribution made a difference. I argue that the proposed metric doesn't do that. For detailed argument: you are just commenting on an answer listing couple of them.
    – yo'
    Nov 28, 2014 at 16:03
  • 1
    @Jaydles - If I post on a 5 year old question, do all of the previous views count towards my "impact"? If so, that seems to be broken; if not, please say so.
    – Travis J
    Nov 28, 2014 at 20:36
  • @TravisJ, it's possible, but pretty rare that it'll matter much - because of the restrictions on what counts for answers, in cases where you're able to get an answer to qualify way after the question is posted, it's very unlikely that the question has many views already. It's important to remember that this number is an indication and doesn't do anything, like say, rep does (and isn't even displayed on your user card, like badges are). Still, we want it to be as decent an estimate as possible which is why we're still looking at ways to tweak it.
    – Jaydles
    Nov 28, 2014 at 20:46
  • 1
    @Jaydles - Following the same example, would it even be an option to only count views after the newly made post?
    – Travis J
    Nov 28, 2014 at 20:51
  • 6
    btw, editing 30 thousand posts? Where do we have the diamond badges?!
    – yo'
    Nov 28, 2014 at 22:56
  • @TravisJ A solution could be to generalize that views come in linearly and only account for views in the appropriate time portion. Though I would only apply such a rule to answers that were posted at least a week after the question (which also prevents the calculation from becoming too expensive).
    – David Mulder
    Nov 30, 2014 at 16:51
16

Since working for Stack Exchange, I've had to explain to a lot of people who don't use our sites what I do. Telling my grandmother about what reputation is doesn't come across so clearly. There are plenty of people out there who (shockingly enough) don't know our sites and don't grok what it is that we're doing here.

I worked for a company that made products that save energy and reduced CO2 emissions. When we presented to people in the industry, saying "Saves 200 tons of CO2/year" means something to them, because they are familiar with CO2 and that raw number means something to them. When the families of employees came to visit though, saying we save 200 tons of CO2/year means nothing to almost all of them. Is that a lot? Is that a little?

So we made a presentation that let them understand what that meant in terms they could understand by saying, "The amount of CO2/year we saved is the amount a forest the size of this town absorbs in a year" -- people can picture the size of their town, which they relate to, and understand that it would take a lot of trees to fill it. The information may be very similar, but the impact differs greatly depending on the presentation.

I see people helped as the equivalent for folks who aren't familiar with Stack Exchange. If I want to stick something on my webpage about my activity on Stack Exchange, this is a great number that means something to a normal person even if they don't know the site. It may even get them to click through and check out the site themselves because it sounds cool.

If I want to stick a link to my SE profile when applying for a job and listing my hobbies, this is a great stat to stick in there, and to have appear if they click through.

If I want to tell my grandmother what the heck it is that my site does, I can point at my profile and say that by asking an answering questions, I've helped a lot of people.

And that is made possible by creating an easy-to-understand metric. It may not be perfect, but it's definitely better than what we have right now in my opinion.

Improve this answer
9
  • 2
    Well said. By that argument, is SE thinking about putting "people helped" in the flair?
    – MTL
    Dec 3, 2014 at 15:02
  • @Shokhet not that I've heard of, but if it turns out to be a good metric, let's suggest.
    – jmac
    Dec 3, 2014 at 15:03
  • You convinced me, i leave aside skepticism, but why not open a question for suggestions on how to improve PH? and let the community to decide how the metric is calculated.
    – rnrneverdies
    Dec 3, 2014 at 17:36
  • 1
    @Shokhet, I think we'd be fairly apprehensive about adding it to anything that suggests it's on par with rep or badges.
    – Jaydles
    Dec 3, 2014 at 18:50
  • @Jaydles I wouldn't suggest that we do it until we're done clarifying exactly what it is and precisely how it's calculated.
    – MTL
    Dec 3, 2014 at 19:33
  • 2
    So... the number is a lie, but it's a useful lie because grandmothers are not likely to know any better. Makes sense.
    – user259867
    Dec 4, 2014 at 0:28
  • @OneWay that's what Jaydles' post is all about
    – jmac
    Dec 4, 2014 at 4:48
  • @Raff, how accurate the number is doesn't change what I see the purpose as. If the purpose makes sense, then it's just a matter of figuring out how to get it accurate enough to work well for that purpose. Tossing out something because it's not perfect doesn't strike me as the ideal solution, so why not work with us to fix the issues you have with accuracy instead?
    – jmac
    Dec 4, 2014 at 4:50
  • 2
    Because the proposal is unsalvageable. Its purpose is to fix European currency system, and the data you have is a chocolate-covered banana.
    – user259867
    Dec 4, 2014 at 5:52
14

We already have the metric denoting "helpfulness" of a person's contributions. It's called reputation. There is no need in introducing anything else because the very essence of gamification (the underpinning of SE's model) is presenting people with an unambiguous criterion of their standing in the community.

Improve this answer
1
  • 1
    Reputation is about the quality of contributions (or at least it's intended to be). This number should simply about what it says it's about (which currently has a problem as outlined at the end of my post~). Either way, although they might sound similar to you, they are very different metrics to some people.
    – David Mulder
    Nov 30, 2014 at 16:38
9

Although there is a lot of backslash against this metric, I actually think it's a fairly nice gimmick if and only if it's only presented on a users profile. Does that change that it's just a gimmick? Nope, not at all, but a lot of people come here to SE because they enjoy helping people, more than they care about reputation or statistics. Showing such a user how many people he helped is something that reinforces that behaviour, which objectively is a good thing even if it tells little about the quality of his contribution (what reputation is all about).

Either way, right now the number is calculated as follow

This is essentially a sum of views of your questions, and answers that fall within the following criteria: accepted, have a score of 5 or more, have more than 20% of total score of all answers on a question or are within the top 3 answers by score. No deleted posts are counted.

Which I think is a pretty solid basis, however a second place answer shouldn't count as much as the first place answer (etc). The number of times I scroll beyond the first answer is already fairly rare, so a ratio 1 to 10 would already seem quite reasonable. Personally I would love SE including a temporary script to observe user scrolling behaviour and actually get official average metrics.

(One problem I see however with this metric is that some SE sites aren't about helping others (at least in their current state). To be specific codegolf.SE and puzzling.SE. Personally I think the real problem is that these sites aren't Q&A sites, but regardless of what the real problem is, this metric would not fit well with these sites.)

PS. Disagree with all of cVplZ answer, except for one detail which I will thus include here, if you're counting views instead of individuals you will need to word it somehow differently. After all, if I help the same person a thousands times that doesn't mean I helped a thousand people. Not sure what the best way would be to approach this though. I actually think the 'number of people helped' metric is far more interesting than the number of times that you helped others. So instead of changing the wording a far more fascinating approach would be to divide the aforementioned number by the average number of pageviews per visitor (a statistic that can be easily found in Google Analytics which SE uses).

Improve this answer
8
  • 6
    Let's call it "views on my Qs and good As", because otherwise the title is very misleading (see Raff's answer).
    – yo'
    Nov 28, 2014 at 10:29
  • 1
    @tohecz: Except then you miss out on the part where it reinforces helping behaviour form those people who care more about helping than metrics.
    – David Mulder
    Nov 28, 2014 at 12:51
  • 1
    You mean the helping behaviour Raff speaks about?
    – yo'
    Nov 28, 2014 at 12:55
  • @tohecz: You might want to check out this network of sites called StackExchange ;-) . That's the kind of helping behaviour I am obviously talking about. You guys might be all addicted to statistics and rankings, but the majority of users isn't. They will just be happy when they see that they have helped n number of users.
    – David Mulder
    Nov 28, 2014 at 12:58
  • 11
    Sorry, I am not. I'm just already annoyed by people doing bad reviews, posting crap, not searching for dupes when they should, and behaving like hyenas just to get some extra points. The proposed system only encourages that. Your answer certainly doesn't address these problems. Once they are addressed, I'm fine. But before that, I 100% agree with Raff. (Just some stats: I've got 480 answers and only 28k rep on TeX.SX, because I care to answer "unpopular" questions. I have 100+ meta posts, 55 of them of score 10+, hundreds of helpful flags and edits. Do I look like a rep hunter to you?)
    – yo'
    Nov 28, 2014 at 13:05
  • @tohecz Seriously? You might want to check the English language... "you guys" is a language construction that inherently has a strong generalizing connotation. It should have been more than obvious what I was talking about in both cases and misinterpreting what I have been saying is just... cheap. And that's still ignoring the word 'might' there which just reinforces that connotation. I obviously didn't check you personally out, but the concerns you were raising where about a certain type of users that I directly addressed.
    – David Mulder
    Nov 30, 2014 at 16:48
  • Either way, to address the comments you're making: You are raising valid concerns about flaws in the reputation system and those should be addressed separately, however this number in no way makes those flaws stronger nor have any flaws been brought up that are newly created by this metric.
    – David Mulder
    Nov 30, 2014 at 16:48
  • 4
    David, either way, I've been active on multiple sites for couple years, so I do know the network. And yes, I raise my concerns, and you still haven't addressed them. And I completely lose you, maybe you use a level of English that is beyond my understanding. To conclude: There are problems I described, your answer ignores them, as I said before. The last couple comments don't seem to change anything on this. If "you guy" are willing to help, why do you need an extra metric for that? If "you guy" are about metrics, you can create dozens of them yourself.
    – yo'
    Dec 1, 2014 at 19:54
8

... and answers that fall within the following criteria: accepted, have a score of 5 or more, have more than 20% of total score of all answers on a question or are within the top 3 answers by score. No deleted posts are counted.

What about the date of the answer?

How a late answer helps earlier views?

The criterion is not sufficiently solid.

Improve this answer
3
  • 4
    Here's a query that returns the highest-viewed questions with fewer than 3 answers. Posting an answer there will be a PH boost no matter how useless it is; as long as it does not get deleted. Expect more of this activity when the new profile rolls out to other sites.
    – user259867
    Nov 30, 2014 at 18:28
  • My understanding was that it counted views since time of answer, because it wouldn't make sense otherwise....but I could be wrong.
    – MTL
    Dec 3, 2014 at 16:56
  • It would be good but not sure of the availability of the date of each view. (perhaps the logs are kept) but at least should be weighted in some way.
    – rnrneverdies
    Dec 3, 2014 at 17:27
5

The "Number of People Helped" statistic needs more transparency. If this goes into effect, then six months from now (let alone years from now), people will be seeing that number and wondering what the heck it means.

While it has been explained here and in other meta threads, these threads will be buried pretty soon, and users shouldn't have to go digging through meta (some don't even realize meta exists, let alone know how to search it) to find what an esoteric stat at the top of their profile page means.

At the very least, a tooltip that explains "This is the number of views your questions have, combined with the number of accepted answers, answers with score of 5 or higher [...]" would be very illuminating.

Fewer clicks to find out what you want to know = success.

Improve this answer
1
  • 2
    Good idea. Maybe also add a link to an explanatory help center page ( then PH would look approximately like a badge )
    – MTL
    Dec 3, 2014 at 16:57

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .