Deleted posts should not influence reputation

Question

A major change was suddenly introduced in the way reputation is calculated yesterday:

if you’ve contributed something worthwhile to the site, you should keep the reputation for that even if it eventually gets deleted. “Worthwhile” here is defined as,

• A score of 3 or greater
• Visible on the site for at least 60 days

This makes no sense: if the answer contribution is worthwhile, why would it be deleted?

Even if the contribution is a positive contribution to the Internet but is off-topic for the site, it should not count in the site's reputation. I may write the greatest ever post about gardening¹, but if I post it on Stack Overflow, it should not count as reputation on Stack Overflow. Because my Stack Overflow reputation is supposed to have some connection with how good I am at programming, not with how good I am at gardening. If you say that a contribution is a contribution and should count, you might as well display a single total reputation figure combining all Stack Exchange sites. That's not what Stack Exchange is about. If we want a questionable unified organization of random answers, we know where to find it.

I can't figure out the justification for the score threshold. I think I see why the visibility threshold was introduced: only posts that have probably had views to match the votes are thus enshrined. The thresholds are an incentive not to clean up bad posts until they meet the threshold. Let my bad posts stay under the radar for at least 60 days, and I'll get rep for them.

The argument for the change is bogus:

In fast-changing professions, there should be no shame in contributing valuable information just because it eventually goes out of date – and there shouldn’t be a penalty for deleting it when it does. Naturally, editing to bring an answer up-to-date is preferable – but if someone else already posted a good answer with current information, you should be able to remove yours and keep the reward for the time it was useful.

No! An answer with obsolete information should not be deleted. It should be edited, ideally to contain up-to-date information, or at least to indicate that the information is out of date. Obsolete information is still valuable. A post about version 3 of a software program does not become irrelevant the day version 4 comes out; it should be updated, if the information does not apply to version 4, but it should stay for the sake of people who still use version 3.

There are a few cases where information about older versions is not directly relevant, but it can still be useful. For example, if a web application has changed its API, a post about the old API won't help people who are programming for the new API, but it can still be a godsend to people who are trying to upgrade an old, poorly-commented application written for the old API. Stack Exchange in general frowns on time-sensitive questions (that's why we have the “too localized” closed reason); that does enough to keep obsolete content out, because we do not allow much content that will become obsolete in the first place.

A lot of Stack Exchange sites (Stack Overflow first, of course, and the pattern repeated on many other sites) went through an early experimental phase where a lot of things were tried. Reputation is an indication of participation based on competence in the field, not on participation in building and maintaining the site; that's why moderation actions and meta posts do not count in participation. The deleted² side of the experimental phase is participation in building the site, it does not belong in the reputation calculation.

To give an example, on Science Fiction & Fantasy, my reputation jumped by almost 1000 overnight, because I had participated in our early “list” questions. After a while, we realized that lists questions were harming the site and purged them; they should not be reckoned in my or any other participants' reputation.

My reputation should reflect my visible posts on the site, same as users who came in later and never had a chance to participate in the old stuff we swept under the rug. Reputation for deleted posts is effectively a bonus for the old-timers. This is not desirable: it tells newcomers that no matter how hard they try, they'll never look as good as the people who jumped on the bandwagon a bit earlier.

TL,DR: **Do not include any deleted posts in reputation calculations.**³

¹ _{Generic I. I am incapable of writing a great post about gardening. Oh, you know what I mean.}
² _{Or should-be-deleted.}
³ _{And more generally — but this is not part of this feature request — don't spring such changes without first discussing them with the community (not even with moderators were informed before the fact).}

Answer 1

I'm not as hard-core a deletionist as Gilles is, but I agree with this 100%: when something gets deleted, rep points should be eliminated as well. For one, because this new policy has a terrible side-effect: it further encourages answering duplicates - even if the dupe gets removed eventually during some purge, people get to keep their rep. That is potentially setting a bad precedent.

At the very very least, the required upvote threshold should be raised massively to 10 or more.

But even with that - as Gilles points out, the policy is likely to favour mainly us veterans. It doesn't set a good example for the site. It will give fuel to people complaining about how newbies will never reach the top rep scores. That argument is largely nonsense, but how can you refute it when there is an elite of users who have thousands of points from popular deleted questions?

This is not a fair change, and it feels like it was designed to assuage some users' whining about lost rep rather than establish a guideline for future high-quality contributions. Please change it. You know I'm all for preserving the cool old stuff somehow, but bending the rules so people get to keep their points is a step too far.

Plus, it replaces what was a simple, straightforward system by a very complicated formula.

Please revert this.

Edit - some make a social argument along the lines of not alienating users that does have a point, especially in light of the tensions over the past few days. I still don't feel comfortable with it, but that's a point of view I can respect. However, I'm leaving this answer unchanged so not to change the meaning of the votes.

Answer 2

I don't think reputation means what you think it means. Reputation doesn't mean you are good at programming, but rather that you've been able to help someone on this site. Reputation is your reward for having taken the time and energy to think about someone's question and produce an answer that at least 3 people found helpful. The fact that it was on the site for 60 days shows that at the time it was likely within the bounds of the site's topic.

Honestly, I'm more concerned that useful, historical information is still be deleted and that leaving the reputation when that happens really serves to hide that fact. The jury is still out, but it's possible that the deletionists have actually won, with the exception of a few notable, historical questions. For example, @casparOne deleted this College Degree question. I know that I found it useful because I up voted at least two of the answers. P.SE doesn't want it, but it was up voted enough that I think it should fit the historical definition. In this case, perhaps, with a link to P.SE indicating that more relevant information can be found there on career aspirations.

Answer 3

No system implemented and maintained by people will ever be perfect.

The argument that rep for old posts that are far off topic shouldn't count toward rep is perfectly logical. Rep should reflect the community's trust in you, and the value of your contributions, and shouldn't be skewed by fun diversions that were accepted a few years ago.

But life is never that simple. I can see how some people would be bothered by being told—"You know those great contributions you made in 2009? The ones we all loved and up-voted highly? Well they're off topic now, and deleted, and we took back the rep from them."

This policy trades a bit of consistency in what rep is and means, in exchange for not alienating parts of our community. I think we all need to remember that it's the community that makes Stack Overflow what it is, and throwing a wrench into that just for the sake of a perfectly consistent rep policy would be bad.

---

Full Disclosure: I'm not an old-timer, and I didn't get any rep back from this new policy. None.

Answer 4

For years now, I've been explaining this reputation system, with the core philosophy of "rep as a measure of experience on the site" coupled with the following behavior: reputation gained from (or lost to) deleted questions will be lost (or regained) once it is deleted... eventually.

And it never made sense.

Oh, it was easy enough to explain. Especially to Stack Overflow users - this is exactly the sort of system a programmer understands. We've slavishly documented all the rules and exceptions in the community FAQ, reverse engineered edge cases, dug up bugs in it...

But every so often, someone would ask the obvious question: "if reputation measures experience, why do I lose it when stuff is deleted? I still have the experience!" And so I used the same rationalization you've used here: if the posts aren't visible on the site, they effectively count for nothing - the experience is meaningless in the context of the site.

But of course in practice it didn't even work the way it was described to work. Reputation recalculation was done on an ad-hoc schedule, usually when something major changed. Individual users might see their reputation recalculated in response to a voting anomaly, suspension, or moderator intervention. Eventually, individuals could request recalcs - but most didn't do so regularly. So you effectively lost nothing when a post of yours was deleted.

Pay attention there: regardless of how we understood the system, explained it to others, and documented it... In practice, it has never worked that way. You kept every last point gained from a deleted question, until months or years later it suddenly disappeared with no apparent rhyme or reason. My rationalization was just that - hand-waving to explain away something that wasn't actually reflected by reality: that number, that "reputation", didn't exactly correspond to anything on the site.

We fixed that. And in the process, dramatically changed the actual reputation system, even while preserving almost exactly that ideal model that was documented but never actually existed for most users.

So what's more important? Maintain the system you and I and a relative handful of other users know intimately? Or the experience that the vast majority of users have known and grown to love over these past few years?

Well, I guess you already know my take on that question.

The exact numbers aren't all that important. We needed a fairly simple formula, and then we played around with the parameters for a while, looking for a balance between rewarding legit participation and preventing easy gaming. There are a few holes that still need to be plugged, but initial observations seem to indicate this hit pretty close to the mark in terms of preserving reputation as a reward for useful effort.

So I hope that answers your implicit question. But there's one point I do need to address:

Reputation for deleted posts is effectively a bonus for the old-timers.

Two months might seem like a long time on a site like SciFi, barely a year out of private beta. If you've only faced the task of removing a single type of question, and are still fully confident that criticisms leveled against another popular bit of fluff are without merit, I can understand how this might seem like it would never benefit that site.

But three years ago, there were folks arguing just as vehemently that career advice was a perfectly good sort of question to ask on Stack Overflow. Some still are.

I don't think for a minute that these sites will ever stop evolving, or that the questions of what is and is not on-topic will be ever perfectly settled. I do know that reputation is a distraction in these discussions. An unnecessary one.

Answer 5

Repution should reflect the status quo, if content gets deleted all the rep should go away with it.

Rep is a fun side effect, but is not why we're here.

Simple is better than complex, there are already enough arbitrary thresholds (at 2k one gets to edit posts, at 10k moderator tools, etc.), let's not add others (60 days and 3 points).

Answer 6

<Disclosure> Recouped 0 on stackoverflow and 1,000 on programmers.SE due to policy change </Disclosure>

While I agree with all your reasons why rep from deleted questions is bad, one very positive benefit with this change is that it removes a barrier to vote to delete bad posts. Including Self-Deletes.

Edit JaredPar describes this well in his feature request Change the way really old questions are deleted

I understand the need to delete bad data from the site. But I think you need to balance the removal of bad data with the negative effects it has on people who are contributing. I propose a small change to this process to help balance the scales a bit.

Answer 7

The system will work best if your score never goes down unless you've done something wrong. Downvoting is so effective because when you start seeing your score drop, you have a strong, psychological desire to correct the problem. But when a question is deleted, you are powerless to fix the problem. It's action at a distance and it's a bad idea.

---

Reputation goes up two ways:

1. The initial increase when people read an answer or question for the first time and agree that it's correct and
2. Over the course of many years as search delivers people who find the answer helpful.

Deleting (or archiving) a question forever eliminates the second type of reputation gain, which is a hidden, but potentially severe punishment. To also eliminate the first type of reputation gain adds (literal) insult to (virtual) injury.

---

By the way, when the site was younger, the first type of reputation gain was potentially much smaller than it is now as more users mean greater potential score. That's because there are so many more users who can upvote posts. That's balanced by the greater amount of time you could be getting upvotes over from the second type of gains.

To be honest, reputation has always been a bit of a crap shoot. Sometimes you are first with a great answer that racks up a ton of points and sometimes you are a little bit late with an equally great answer that is ignored. Some of the posts I am most proud of (on all SE sites) have languished with +1 or something like that and may never be recognized as the brilliant insights I know they are. Then I answer a Perl FAQ and "earn" +34 on the third best answer out of three. It's humbling sometimes.

Answer 8

Well, I'm pretty certain I wrote a comment here, but it appears to have disappeared, so here it is again, from memory:

This doesn't make any sense to me. If that reputation is worth something, it means that the associated post was worth something, so why delete it?

To me, this seems like applying the wrong solution to the wrong problem.

I think you should stop implementing this right now, and work on Pekka's idea of 'archiving' (or some variant of it).

Answer 9

Reputation is meaningless. With the recent grandfathering changes, they've managed to make it even more meaningless.

Now the flimsy bar you all self-evaluate each other on about who has more programming chops has even less basis in reality with power ties, wowing people in interviews and marzipan over straight frosting into the frothy mixture.

{Topic du jour and adjunct chatter about spandex, university courses and plastic bagging as analogous comparison for other SE sites.}

There is a truth undeniable that you can't escape the EXP, they will never let you give it up or let it go. It is definitely not going to desert you.

Reputation is the reward you get for clinging on for dear life and obfuscating the historical archives in guise of preserving content. Them riches are the reputation and you must placate the user base in their inflation.

You don't rack millionaires by comparing them against people late to the party.

Answer 10

Full Disclosure: I lost about 300 points when the correction happened a couple weeks ago. I then just this morning saw that I gained all/most of it back. I know several times I deleted my posts when (a.) I later found them to be partially incorrect or otherwise inferior, but (b) they had garnered a lot of votes. I'd be worried the OP would go with my answer, and I'm a little embarrassed to be the top-voted answer when it's clearly (a little later) not the best. If I later found my answer to be inferior but had only 0 or 1 upvotes, why bother deleting? It was already getting the votes it deserved! Not sure if this accounted for all my points: Maybe 1 or 2 of my other answers were considered dups and deleted by a moderator.

I certainly agree with Gilles about not deleting "outdated" material. That should be kept indefinitely if not forever. If a would-be archivist wants to improve the tags to make clear the previous version or tooling for which it's relevant, then they are doing a great service. I also agree that the 3-or-more cutoff and been-around-a-while is not a "balance" that helps.

As for duplicates, I would suggest the poster be given half-value (5 points per upvote) for any gained rep, no matter how much. This is a balance that I believe would motivate people to take the most-desired behavior. It certainly would with me. I believe myself to be a typical somewhat-selfish, yet somewhat community-oriented and helpful stack-overflow user.

I deleted my own posts at first because I found myself to have really misunderstood the question, yet had gotten votes. (The Disciplined badge helps that first time of course.) Like making your first kill, it was easier later, but always somewhat painful because I knew the question would, if left, get me more rep. Once I saw that it didn't seem deleting my posts cost me the rep already gained, it was certainly easier to do! It cost me a little, but not a lot, and it was the right thing to do, so I did it. If I lost all my rep for deleting my own post, hmmmm, well, I would do it if the situation was really bad, I guess, maybe, but I'd be a lot slower to do it. If I was going to keep half however, I'd certainly want to be a good citizen.

And if I got half-points for an answer to a duplicate question, that would eventually be deleted, I'd (a.) not be really pissed off if it disappeared later. And you know, I did really help that guy, and I should get some credit for that. But if I only got half-points, I might be more motivated to not jump in with an answer, but do the research to close it as a duplicate. That's the main thing we're after, right? We want to decrease motivation to answer duplicates, to answer off-topic questions, but still give some credit for helping people quickly and effectively, and yet not completely punishing people who mistakenly answered duplicates.

Imagine you're a new user, answering questions you can, all excited, and then you get caught in a crackdown on duplicates and/or off topics, and you see your points drop from a hard-earned 180 to 70. You are pissed off and de-motivated and not having fun. Well, if it instead dropped down to 125, your not as pissed off at least, and that's good. Bottom Line: If someone quickly and effectively helps the OP, and is recognized for doing so, completely removing their rep later is a bad move. Reducing it, on technicalities, is not nearly as bad, is fair.

It's balance. Is losing 1 rep for downvote fair? No, but it's good balance. Is 10-for-upvote but 2-for-downvote fair? No, but it's good balance. Partial credit for deleted content is good balance.

Answer 11

I think I agree that on the main sites, deleting means losing (or gaining) rep, but on meta, especially this meta, deleting old bug reports should not change the reputation gained from finding them.

Answer 12

Old timer chiming in...

Reputation is a number. It used to be referred to as trust in the system. Someone with 200 rep on META may not seem like someone who has been around as someone with 75K rep (apply same pattern to all sites). Reputation is also a reward for effort. While we answer questions to try and be helpful, once someone decides the question is "off topic" or "not constructive" it gets deleted. Never mind the fact that the OP had a real question that had a real answer.

A great example of a real question that provides value to stackoverflow was Joel's "How do you move a turtle in logo?". I can't seem to find this question now which leads me to believe it is long gone. Was Joel proving a point with his question? You bet. But is there a programmer out there somewhere who may have sought that information? You bet. The last time I looked at stats for web traffic to SO, it was clear that direct traffic was a very small percentage meaning the site received hits from search engine traffic. So why remove questions that provide hits to the site?

I actually think there is a better analogy for what is really happening here... In the US there is a bit of class warfare going on. Whether it is the rich oppressing the poor or politicians trying to steal from the rich to feed the poor, having money makes you a target. I believe that folks in their 70's have had the best chance at acquiring vast amounts of wealth because of the opportunities that they've had in life. These folks in their 70's are probably similar in nature to those "old timers" on stackoverflow. Now these younger generations in today's world want everything their parents and grandparents had, and they don't want to wait for it. They also don't really want to work for it. Instead they try to point out the injustices in the system because you used to be able to do X, Y, and Z to quickly make money and you can't anymore... I think you see where I'm going with this.

At the end of the day, there are undoubtedly useful questions on this site that are being deleted. Maybe as the FAQ and guidelines have evolved, these questions fall out of favor... But I still don't see the point in deleting them. What injustice do they serve? Who is paying for them? The last time I checked, StackOverflow is FREE to you and I. As a business, StackExchange needs to make X amount of dollars and removing questions that provide search engine traffic just doesn't seem right. I know that Jeff and Joel have gone out of there way to try and let the community self manage some of this stuff, but behind all of this is a business.

Ultimately I see all these deletes as being a negative for the business, for people on the web seeking help, and for all the people who contributed content to those questions and answers in the first place. The only injustice that seems to have happened here is having users lose reputation (their reward) after they spent the time to be helpful. There are no other injustices out there. If you want rep today, operate under today's FAQ rules.