3

With or without its will, Stack Overflow is attracting a lot of attention by the scientific community. See for instance the lengthy list of research articles at Academic papers using Stack Exchange data. Some of these papers already target sociological questions, and I expect more research to emerge in this direction, to study cultural contexts, ethnicity, gender inequality, etc.

Among the pieces of data useful for such studies are the profile pictures of Stack Overflow users. For example, one could try to infer gender, ethnicity, or aspects of cultural self-identification by inspecting them.

However, even though email hashes are part of the data dump already for some time and could be used as Gravatar links, there are users for which their email hash does not match their Gravatar hash. I think the mismatch has something to do with the way one logs into SO, e.g., using Facebook, but feel free to correct me if I'm wrong.

Moreover, there are users with pictures who don't use Gravatar at all, but use something else instead, such as a Facebook graph.

Therefore, if interested in the pictures of Stack Overflow users, the only way to obtain these links now is by scraping the user pages, which is cumbersome, slow, creates unnecessary traffic, etc. Since I thought the entire idea of a public data dump is to support studies of Stack Overflow rather than functioning of Stack Overflow on its own, I was wondering if it is possible to include links to the profile pictures in the data dump (e.g., as a separate entry in users.xml).

Including Gravatar hashes has already been suggested in an earlier post from 2009, which is still unanswered. However, as described above, this would not suffice.

Improve this question
19
  • 4
    Ok, so we've got an opinion. What's missing is the "problem", and preferably a narrative that gives them a reason to do what you're asking. Who cares if the user-hash is or is not the gravatar hash? You've convinced me to not care about your request, by not making it useful.
    – jcolebrand
    Aug 7, 2012 at 14:10
  • 1
    @jcolebrand, thank you very much for the comment. I've edited the question to try to make it clearer why this would be useful, and who could benefit. I hope it helps.
    – Bogdan Vasilescu
    Aug 7, 2012 at 14:52
  • How does one utilize the pictures shown on the profile of a member to determine anything about the cultural context of the user? I might as well just set my face to this one: gif.neuralcorrelate.com/finalists_2009/Russell/… and proceed as normal.
    – jcolebrand
    Aug 7, 2012 at 14:54
  • 1
    Any 'data' you would get from direct links to the images would be extremely subjective. Certainly, any scientist should have major reservations about using those images for anything at all. It would be like taking a poll by asking, "say whatever you want", and trying to infer respondents' color preferences from the results.
    – Andrew Barber
    Aug 7, 2012 at 14:59
  • 3
    Looks like what you're doing directly contravenes the gravatar TOS: "do not distribute, display, perform, use or otherwise exercise or exploit any User Submission or other user information for any purpose other than as necessary to perform the Services for the applicable user; ".
    – Flexo
    Aug 7, 2012 at 15:08
  • @jcolebrand, by cultural context we mean references to movies, books, political events etc common in the community (or parts of it). For instance, numerous SO users make use of the Guy Fawkes mask popularized by the "V for Vendetta", SouthPark, IronMan, The Big Bang Theory and SpongeBob SquarePants. The first reference seems to be particularly interesting: David Lloyd stated "the Guy Fawkes mask has now become a common brand and a convenient placard to use in protest against tyranny".
    – Bogdan Vasilescu
    Aug 7, 2012 at 15:10
  • 1
    Thanks Andrew, glad I'm not the only one seeing an issue with this request. It doesn't seem very objective at all, and subjective data studies are often unscientific. Additionally, Flexo makes a great point, and this is something we in this community can help you sort out, if you're going against the TOS of a provider. I for one don't believe this study would have merit, and would be open to an NDA and complete disclosure discussion to see the benefits, but I am not at liberty to discuss anything on behalf of Stack Exchange Inc.
    – jcolebrand
    Aug 7, 2012 at 15:20
  • 2
    @AndrewBarber Comparable studies of Facebook/dating sites' images have been conducted in the past: indeed profile images can be seen as a form of "implicit" identity construction [astro.temple.edu/~bzhao001/…. Previous studies have shown differences in self-presentation between women and men, as well as between different age groups. For instance, on a dating cite women are more likely to be shown smiling or to emphasize their sexuality [cyberpsychology.eu/view.php?cisloclanku=2008110402].
    – Alexander Serebrenik
    Aug 7, 2012 at 15:35
  • But Alex, this site doesn't give the users a motivation to adequately represent themselves, and you suffer much more self-selection bias here than on a dating site. You're comparing apples and pears man. Both are green, grow on a tree, have juicy fruit, but they aren't the same.
    – jcolebrand
    Aug 7, 2012 at 15:45
  • @Flexo Thanks for pointing this out. I have to think about it. In any case, it is not yet clear to me what "exploit" means.
    – Bogdan Vasilescu
    Aug 7, 2012 at 15:45
  • As of the top 20 users on [meta] of the default ranking, you see 2 gravatars, at best 6 human faces, of which two are partially edited, so you've got 1 in 5 to judge faces with, 1 in 10 are gravatars (nothing to be gleaned from this information) and the others are not necessarily cultural icons. i.imgur.com/mp8gM.png I'm not entirely sure what the data is supposed to yield, so since this seems like a busy-work request, I'm still of the mind this is downvote and close worthy.
    – jcolebrand
    Aug 7, 2012 at 15:49
  • @jcolebrand: I like you observation that this site does not give the users a motivation to adequately represent themselves. BTW neither does the dating site: it motivates participants to represent themselves in a "better" way (better = younger, sexier, ...) Does this also mean that SO also motivates users to represent themselves in a "better" way with "better" meaning "nerdier/geeker"? Would somebody put a regular photo on FB and a sci-fi/BigBang image here? Many questions can be asked ;-) Self-selection bias is, of course, always an issue but there are ways to deal with it.
    – Alexander Serebrenik
    Aug 7, 2012 at 15:52
  • @jcolebrand One of these images represents a ninja (= expert onlineslangdictionary.com/meaning-definition-of/ninja), i.e., this image can be seen as related to identity construction.
    – Alexander Serebrenik
    Aug 7, 2012 at 16:03
  • Oh I grok that, I just don't think that the gravatar hashes here will be most effective for that. Especially given the couple hundred thousand (minimum) SE users, and how they choose different images to put up across sites, or not at all. And also the historical information for those has been lost. I know of one user who historically uses a gravatar, but on occasion updates it to an image. Then reverts to a gravatar some short time later. His choice, of course.
    – jcolebrand
    Aug 7, 2012 at 16:25
  • 1
    Ok so we contacted Gravatar about this and this is what they said: "That should be fine - the TOS is more for if you're using a Gravatar account, or if you're using our developer API. As far as I can tell, you're just gathering publicly available information. If you are using the API we do have a TOS point that says you're not allowed to store or modify user submissions. Best wishes with your project!"
    – Bogdan Vasilescu
    Aug 8, 2012 at 20:00

1 Answer 1

Reset to default
2

I'm answering this question:

Therefore, if interested in the pictures of Stack Overflow users, the only way to obtain these links now is by scraping the user pages, ... I was wondering if it is possible to include links to the profile pictures in the data dump (e.g., as a separate entry in users.xml)

The users table / users.xml now contains an attribute ProfileImageUrl with for a majority of users an URL to the users profile picture. For yet unclear reasons not all ProfileImageUrl's are populated:

click for sede

but enough are so I guess that could feed the academic curiosity for now.

Improve this answer

You must log in to answer this question.

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