8

The "last seen" information is on every user profile, and I would argue that in some cases it compromises user privacy and voting anonymity (though others clearly disagree). I think this is particularly an issue on smaller sites in the SE network.

Why does the "last seen" time need to be so precise? I'm fine with "last seen 2 hours ago", but how is it useful to know someone was "last seen 2 minutes ago"? It is useful sometimes to know if a user has been on the site in the past day, or perhaps the past hour, but I cannot think of a reason why anyone needs to know another user has been on the site in the past 2 minutes.

Furthermore, the last seen time is inaccurately precise. The time is not updated instantaneously, meaning that when it says "last seen 2 minutes ago" someone may have been active more or less recently. Which means that the site is displaying incorrect information some, maybe most, of the time.

I therefore propose that "last seen x minutes ago" be replaced with "last seen <1 hour ago" or similar. I feel this would provide just as much useful information, while providing better anonymity of site use and voting, and also more accurately represent user activity.

  • 11
    It's not that precise. That time only updates once every 15 minutes, so they could have been active for the entire past 15 minutes but the site could still say last seen 15 minutes ago. Guessing that someone voted based on the last seen time is an incredibly inaccurate guess. – animuson Oct 1 '16 at 19:15
  • 2
    I am well aware that the last seen time is not updated instantaneously, but based on my experience on small SE sites I strongly disagree that it is an "incredibly inaccurate guess". – Harry Vervet Oct 1 '16 at 19:23
  • 3
    You would have to go through all profiles to check the "last seen", then aggregate in 15 minute buckets (when do you start? At how many minutes past the quarter?) - your "suspects" are the people within a bucket. Even on a small site, this will be a number of people - meaning you can't reliably get any conclusions about any one of them. – Oded Oct 1 '16 at 19:26
  • 2
    I don't buy this. Is the site really small that only one frequent voter is active at a time? – M.A.R. Oct 1 '16 at 19:26
  • as far as I can tell voting anonymity doesn't leak this way but still, I would be interested to see an explanation in the answer to this question. And no, vague "system takes care of that but details are secret" wouldn't do. Anonymous votes don't look like the kind of thing where security by obscurity suffices – gnat Oct 1 '16 at 19:52
  • 2
    Given the downvotes, I have reworked this question to be less about voting and more about the proposal – Harry Vervet Oct 1 '16 at 19:59
  • 3
    The issue on de-anonymising voters seems to be unsupported, but I am also in favour of an hour-based accuracy, not a minute-based accuracy. Updating in 15-minute blocks is a great method for putting coarseness into the data but the resulting data should be at least as coarse too. – Nij Oct 1 '16 at 22:29
  • 2
    As another regular user of the Genealogy & Family History SE I agree that it seems relatively easy to guess who has downvoted on our site sometimes, and that "last seen <1 hour ago" would be likely to reduce the confidence that I have in some of my guesses, which would be a good thing. – PolyGeo Oct 2 '16 at 3:23
  • 1
    Are down voters hunted down on that site @PolyGeo ? – rene Oct 2 '16 at 15:12
  • 1
    @rene Not at all, in fact we don't get enough downvotes sometimes. But anything that potentially undermines voting anonymity should be reviewed. In any case, as I stated in a comment above, this post is not only about voting. – Harry Vervet Oct 2 '16 at 15:43
  • @HarryVervet I still have my down vote on this question despite your effort to clarify. I still get the impression that based on assumptions and hearsay you conclude anonymity is breached and because of that users tend to hold back on their actions. Do you have an idea of how many users hesitate in using the site for reasons of those anonymity concerns? – rene Oct 2 '16 at 16:51
  • 1
    @rene No, I of course cannot possibly know this – Harry Vervet Oct 2 '16 at 17:04
  • @rene I've seen no evidence of any "hunting down" or payback voting on G&FH SE which I think has a very mature, robust and respectful dialog between its users. Guessing who may have voted in particular cases is not an issue, as far as I know, due to the relative anonymity of most votes. – PolyGeo Oct 2 '16 at 19:51
  • @animuson Going back to your original comment, just looking at a user's profile and refreshing the page every few seconds, I saw the last seen time change from "just now" to "10 seconds ago" to "13 seconds ago" to "18 seconds ago", etc. I therefore don't buy your comment that it is updated only once every 15 minutes. On the contrary, the last seen time seems to be very precise. – Harry Vervet Apr 30 '17 at 1:26
  • @HarryVervet You misinterpret what I said. The timestamp for when they were last seen is only updaed once every 15 minutes minimum. That's how the system works. What you're observing is merely how long ago that timestamp occurred. Of course that would change. – animuson Apr 30 '17 at 1:29
7

I am in favor of your proposal to make it a little harder to guess who voted, although it is almost impossible to do so now. I don't think you would go over the user profile of all users in a site (no matter how small, it would be quite a job to visit them all) and just find one user online at the time being able to vote.

I think the most important argument in favor of this request is to stop users thinking they can guess who voted on them based on this single number. They can't say that any more if it just reads seen this hour.

One argument against this feature request is that it is a useful indicator whether a user who asked a question has actually seen your comment or stays with his question or not.

-3

I think it is obviously a no-problem on large sites.

It is also a very minimal problem on the small ones - I can't imagine such a small site where really a single user had visit it in a minute. Note, how can you search for the "last seen" field? I think, it may be an anonymity problem if the OP has already some suspection from the possible downvoters.

Checking their "last seen" field, he gets only a stronger or weaken suspection, possibly in the false direction.

I don't think it would be a problem. The solution against the legal downvotes is that you write enough good posts.

Against the illegal ones, well, there is no clear defense, but their possibilities are significantly hardened by various precautions of the SE system. I don't think they could cause real harm, except a little psychological effect. Maybe it would be useful to make clear for the users, it is their task to not let themselfes be affected by downvoting schemes.

Trying to find the downvoters by indirect means is bad solution with the high chance of misleading. Users trying to do this cause more harm for themselfes on the long-term as to the suspected downvoters, and in my opinion, it doesn't really depend on the precision of the "last seen" field.

You must log in to answer this question.

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