18

Every once in a while, when someone asks a question, there is a flurry of attempts to answer the question without referring to any supporting historical evidence. At best, such answers are purely speculative; at worst, they address an entirely different question. This is especially true of questions, i.e. those asking about the rationale of some given design at the time it was created, where people often instead opine on whether the design makes sense to them personally, or try to guess the rationale based on anachronistic considerations.

Take for example:

  • How did the IEC decide to create kibibytes? asks what rationale the IEC had to devise a new set of unit prefixes based on powers of 1024. Many answerers just wrote on whether they personally find separate set of prefixes sensible. At best disk manufacturers were blamed for the 1000/1024 confusion, without any sources given. The most authoritative answer was provided by Stephen Kitt, though ironically, in a comment.

    (Meanwhile, the quite similar question Could we have avoided the whole UTF-16 fiasco? has been closed as opinion-based, even though, ironically, the answers under that one are in fact much better grounded in actual history.)

  • Under Why was `!` chosen for negation?, we have an answer attempt based on ‘emotional interpretation’, another (unsourced and, it seems, ahistorical) suggestion the symbol was borrowed from ‘propositional logic’, and another one that for example discounts | based on confusion with Unix pipes, ignoring the fact that Unix and pipes weren’t invented until much later.

(I may add more examples later.)

I guess, with computing being a quasi-naturalistic/-philosophical/-empirical domain, where a ‘why don’t you try it yourself’ attitude is common, there is a tendency to assume that the answer to every question may be derived from first principles or discovered by a simple experiment. History doesn’t work that way, though. Either your claims are based on a permanent record of what happened (which you may need to interpret to one degree or another, but you still need to have), or they are unfounded: it’s that simple.

(I can sometimes begrudgingly accept personal recollections of what happened: as faulty as human memory might be, it still beats trying to re-derive history from first principles.)

Personally, I just downvote such answers on the spot (and even vote to delete them when possible), but I feel this might be not enough, given that they also tend to receive a number of sympathy/superficial upvotes as well, especially when the question ends up in Hot Network Questions.

What else can we do to have fewer answers of this kind?

19
  • 4
    I don't know but it's an excellent point well made.
    – TonyM
    Jun 23 at 11:24
  • Unless you know the correct answer, your downvotes are also just a matter of personal opinion. The obvious way to discourage it is to ban all users with a rep of less than say 5,000 from answering any questions at all. Oh, and remove any rep points gained by voting for answers which are not "accepted" to stop people gaming the rep system :)
    – alephzero
    Jun 23 at 12:54
  • 4
    @alephzero, Unless you know the correct answer, your downvotes are also just a matter of personal opinion That's not at all true, there's not an equality of the two there. One doesn't have to know the exact answer to something to spot a wrong answer. Example [Q: When was the mouse invented? A: 1997 Reason: I was using one 10 years before - I know it's wrong but I don't know the right answer] and many millions more. Otherwise multiple choice questions would have no value.
    – TonyM
    Jun 23 at 13:14
  • Interesting that the ‘emotional interpretation’ answer for the 'Why was ! chosen for negation?' has more upvotes than downvotes...
    – UncleBod
    Jun 23 at 13:29
  • 1
    You could word the question so that it explicitly requests answers with sources. That's what I did when I asked about a weird Fortran feature recently.
    – texdr.aft
    Jun 23 at 23:12
  • 1
    Speaking as an armchair historian who occasionally comes up with actual references, I'm not sure why we want to discourage this. We're not professional historians. I enjoy the back-and-forth with people who have different views of the birth of whatever the hell it is that I get paid for doing every day. Sometimes if it's my question, it's exasperating that all you heathens wander from the point, but I can live with that. Even when it's clear that the only person who knows was the person who decided to use "!". The discussion, not the answer, is the point for me. Jun 27 at 0:10
  • 1
    @another-dave We certainly shouldn't forbid conjecture or speculation entirely. I like DrSheldon's proposal, since it means that only questions with the reputable-sources tag will be strictly moderated, not all history questions in general. Discussion would remain perfectly fine in answers to non-reputable-sources questions and in comments.
    – texdr.aft
    Jun 27 at 3:25
  • One prerequisite for this would be for the SE site in question to generally accept that such questions would be left open and available for months on end until a primary source or participant were able to respond. Some SEs really don't like questions being open when an answer isn't immediately forthcoming.
    – knol
    Jun 27 at 5:42
  • Indeed I wouldn’t banish speculation completely, but it still should be considered a last resort, when it’s clear that all other routes have been exhausted, and clearly marked as such. Jun 27 at 7:48
  • 2
    @texdr.aft - I'm not so sure about that; it grates my anti-authoritarian side, in that it allows the questioner (or indeed any other tagger) to control what is "allowed" in responses. I prefer the current system where the questioner can indicate preferences, which wish we can then politely follow or impolitely ignore. Jun 27 at 12:31
  • 1
    Well, if we want this site to become as hostile as Stack Overflow.... Jun 28 at 1:50
  • 1
    This site seems to have worked pretty well for the few years I've contributed here. It has been loosely policed. Most of the interesting results have come to light by back-and-forth discussion between people who have different perspectives on historical events. With respect to the 'hostility' - SO insists it is a Q+A site for "professional" programmers despite the self-evident truth that it's a site for questions from raw beginners, and this results in exessive zeal for closing questions on the flimsiest pretence. I don't want RC to go the same way. Jun 28 at 12:55
  • 1
    @another-dave, this question's made you imagine far worse than the specifics the question's about. Ironically, it means you're writing answer-comments that don't address the actual question - what the question's trying to stop :-) Ease down, no-one's trying to stop the site being pleasant. Just to see the back of that minority of rambling or long-winded that we'd try to get away from if these were actual face-to-face chats in a pub, back in the days when we were allowed out. Posts should be useful, they're not just to please the poster. Again, it's not a forum - it's a Q&A site. Always was.
    – TonyM
    Jun 28 at 19:54
  • 1
    @TonyM ‘appallingly long, dull essay answers’ – I feel low-key called out by this. (Especially that I’d kind of agree; I’m somewhat dissatisfied how verbose my answers tend to turn out.) Jun 29 at 11:10
  • 2
    @user3840170, hey, you're quoting me right out of context - I'd written "some appallingly long, dull essay answers that it's painful scooping out bits of answer from chatty ramble". I also enjoy long answers that are suitable for the question and make a progress through their point - they're great and there's plenty here. It's those where I have to repeatedly sift through a lengthy text that jumps around and try and decode it to answer - they fail in their basic job: to answer the question clearly and readably. I'm not objecting to answer lengths there but to answer quality :-D
    – TonyM
    Jun 29 at 15:28
10

Sigh. Some of my own questions...

Did the Apollo Guidance Computer really use 60% of the world's ICs in 1963?

Please support your answer with references, rather than speculating.

Origin of "arithmetic" and "logical" for signed and unsigned shifts

Please support answers with sources, rather than speculating.

Why were relays prevalent in early 1940s computers when vacuum tubes were also available?

"Because that was what was available" may possibly be an answer, but it needs to be supported by a reputable source, rather than speculation or argument.

...and yet most answers were pure speculation.


My suggestions:

  1. Establish a new tag:

    The person asking a question wants an answer supported by a reputable source. All answers must include relevant quotes, citations, or links. Answers based solely on opinion, argument, or speculation are subject to deletion. This tag should never be the only tag on a question because this tag frames how a question should be answered, not the topic. Please flag the moderators to add a warning banner to the question.

    (If you have a better name or description, put it in the comments below.) Then the asker can choose if they want this. Even better if we can make it one of those alternate-color tags. The idea is similar to the hard-science tag in Worldbuilding.

  2. Create a banner that moderators can add to the top of such questions:

    This questions asks for reputable sources. All answers should be backed up by relevant quotes, citations, or links. Answers based solely on opinion, argument, or speculation are subject to deletion.

  3. (Optional) Allow an exception to the policy that speculative answers can be posted in a community wiki. Not sure if this should be a general exception to the policy, or done on an individual question basis (e.g. different tags).

  4. Aggressively delete answers that violate the above. This is to be done by the usual flagging process and review queues. Possibly add a custom reason for deletion.

  5. Hold a vote here in meta approving such a policy. This question itself doesn't count, as we are in a discussion phase.


It's worth noting that Monica Cellio (yes, that Monica) came up with Worldbuilding's hard-science tag.

Worldbuilding's policy on deleting answers that don't comply with the hard-science tag is here. Note that the top-rated meta answer is basically to leave it up to the community to police answers. In practice, non-complying answers are given a warning and time to correct the problem. Deletion is only done as a last resort, using the "low quality" reason. There is very little work for the moderators (other than adding the question banner). There is no "autopilot" that automatically deletes answers.

Read Monica's meta answer, and you will see that she also deleted answers for not providing supporting information.

37
  • 1
    It's a proposal to change one of the most basic policies that make SE tick.
    – tofro
    Jun 27 at 12:05
  • 1
    Now you have me baffled. How can worldbuilding have incorrect answers?
    – tofro
    Jun 27 at 12:18
  • 2
    Can I add the 'reputable sources' tag to someone else's question? Can that cause the autopilot to delete existing answers? (Don't mind me, I'm a programmer; I'm supposed to think through the possible consequences of making changes). Jun 27 at 15:46
  • 1
    @MaciejStachowski: Wrong. Just try posting a non-sourced answer to Skeptics, and see how fast it is deleted. And English does require answers to single-word-requests to be supported by a definition or quote from somewhere else. It's strange that you think requiring sources is unreasonable for a Q&A site, and that writing an answer out of your behind is okay.
    – DrSheldon
    Jul 2 at 14:04
  • 1
    @DrSheldon I do find the idea of requiring the reputable-sources tag somewhat bizarre. To me being able to source your answers with at least secondary sources should be the default, and anything else should be either opt-in, or closed as opinion-based. (Doubly so for "justification" questions, where I could come up with 20 potential reasons for each of them, none of them valid). Jul 2 at 15:06
  • 2
    @MaciejStachowski - the comparison to "making up words" is extreme. I don't think we're talking here about raw invention but about informed speculation. The nature of RC is that personal recollection about events of 50 years ago often gets to the heart of the answer. Jul 4 at 14:01
  • 1
    @another-dave: No, reputation is strongly correlated to how early the answer was posted, not the reliability of the information in the answer.
    – DrSheldon
    Jul 4 at 17:47
  • 1
    @MaciejStachowski Many of those who contribute to RC seem to disagree. And we have personal knowledge of the history, being there and then. Jul 4 at 20:46
  • 1
    @MaciejStachowski, what you don't seem to appreciate is that what you think are citations in reliable sources are open to all the same criticisms that you are throwing around here. Even a peer reviewed academic paper can easily offer an unbalanced view or personally coloured judgements. Ultimately, the quality of responses is not decided by the quality of citations people choose to employ. Esp. when questions are forcing opinions that cannot possibly be justified by a reference, because, believe it or not, not everything is known, decided and available on the internet, or even in print.
    – introspec
    Jul 5 at 0:37
  • 2
    @introspec: Even if a true answer is not available, that does not justify pretending that someone's guess is a true answer, nor should it preclude an asker from trying to get the true answer.
    – DrSheldon
    Jul 5 at 1:06
  • 1
    @introspec There's reliable sources, unreliable sources, anecdotes from anonymous people on SE, and speculation. And some questions are not answerable with a reference, but that's fine. If I ask "how were Windows logo colors chosen", I'd expect an account from a Microsoft engineer (or a contemporary article, etc.), not an answer from someone at SE purporting to have worked there or someone just speculating they went with primary colors. Other sites distinguish between original research and references just fine, why can't RC.SE? Jul 5 at 7:15
  • 1
    @MaciejStachowski, other sites, e.g. Wikipedia, on some occasions threw out original content by top people. An account from a Microsoft engineer won't cut it at Wikipedia, because they don't accept personal accounts. At the same time, there are many engineers, incl. from Microsoft, who shared their personal accounts at this very website. What I am trying to show is that references do not magically appear out of thin air. Someone has to do an original research. And prohibiting original research in the articles is insanely misguided. Yes, not everyone can do it well. We have voting for that.
    – introspec
    Jul 5 at 9:00
  • 1
    @DrSheldon, every piece of research starts as someone's guess. Who made you a judge of its quality? To Illustrate, I have had a look at your Qs and As. Your top question is about the difference between . and -> in C. You gave yourself a tick for the top answer, seemingly because you found a citation from K&R that explains why both operators are necessary. At the same time, you didn't accept a higher ranked and earlier answer by supercat, which states the same thing very succinctly and mentions the same original source.
    – introspec
    Jul 5 at 9:09
  • 1
    My real point is: in any field of human activity, there are differing opinions, however close the people were to the source of original informations. It is OK. Good answers on this website use all sorts of approaches to come up with answers that they believe are representitative. Some answers are not good. Some answers are too shallow or not grounded in enough facts. However, what makes them bad is not the fact that they are too personal or opinionated, or that they do not include enough citations. They are bad because they do not reflect the current understanding on the specific topic.
    – introspec
    Jul 5 at 9:20
  • 1
    @introspec, I think you're arguing over the fine detail of the exact border. The guess-answers I objected to are obviously far over any line. Posters know when they've invented/'deduced' an answer, rather than seen evidence of their answer or are relating established fact. Go back to basics: this is a Q&A site, not a discussion forum. So venturing unsubstantiated opinions, no matter how much fun it is (which is most of the trouble), is the problem. Readers might pass off what they see here as fact, breeding disinformation. That's a common internet problem but let's not deliberately add to it.
    – TonyM
    Jul 5 at 12:15
2

I think this type of questions is asking for it. They are just opinion-based questions in disguise. Of course, we can probably think of examples of questions of this kind where somehow (miraculously) a well-sourced answer has actually been given. However, frankly, if you really ask for explanation "How did the IEC decide to create kibibytes", you've got it coming, because even the explanation from IEC website linked to by Stephen Kitt is nothing more than some plausible hand waving.

So, overall, for specific situations where poster expects for some reason that precise answer must exist, DrSheldon's protocol of enforcing sourced answers may be helpful. However, I would be very wary of extending his suggested mode of moderation too widely. The whole idea that reliable information must be supported by "reliable sources" is a bane of Wikipedia and it would be wise to avoid implementing something similar here. Many specific questions on this site were answered by running a program, or looking into an open code, or by disassembling some old binary. None of these examples fit the narrow Wikipedia-like definition of "reliable source". The possibility of presenting your own research is what makes StackExchange really helpful.

Ultimately, the real problem that you are trying to solve is that voting disproportionally benefits quick answers, not well-researched answers. Well-researched answers given late are often treated as irrelevant, esp. for questions with many answers. So people are actively discouraged of doing any serious research if the question already has 5-10 answers, because they know their research won't be read anymore. If you can address this dynamic, many of your concerns will be addressed automatically.

3
  • My proposal leaves it up to the question asker. If the tag is not added to a question, then speculative answers are allowed. It seems to work well for Worldbuilding. Also, I completely agree that fast answers are unfairly rewarded more than well-researched answers. An answer that is well-written but a day late ends up below poorer-quality answers, and has a hard time rising to the top.
    – DrSheldon
    Jun 27 at 11:58
  • Totally agree that some of these "history" questions really provoke (sometimes, even more or lesss obvious, but still) speculative answers. More often than not, it's the question style that's bad, not the answer.
    – tofro
    Jun 27 at 12:00
  • Three downvotes and not a single comment to express your actual opinion? This site used to be a lot more transparent.
    – introspec
    Jul 5 at 9:23

You must log in to answer this question.

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