How should TMW repair this answers flaw?

The problem I see is that new users don't understand the answers interface. Why do I say that? Because when a new user asks a question and get an answer, they often want to ask a followup question, which SHOULD be asked as a comment on the answer. This is entirely ok of course, and encouraged. It shows that a person is interested in the answer, and wants to learn more.

The flaw is, how should they do that? What do they see?

The new user sees a LARGE box, for a new answer. There is also a tiny blue button to add a comment to that answer. But they see a LARGE BOX to type in.

Honestly, what will the novice user type in to ask a followup question or make a comment? The big BOX of course, that lets them type. So every new user keeps adding new answers to their question for every comment. Eventually they learn, but only if someone is able to get their attention.

In one case, I saw Image Analyst actually move an answer into a comment, deleting the answer and pasting in the text as a comment in the appropriate place. He then added extra text to explain what he had done and why. I considered doing it myself at least once, but this takes extra work, and given the number of times it happens it is clear that this won't be done often by someone who has the powers to make that move.

My point is, the problem is the choice of user interface for a response. Big visible edit box for text (as an answer) versus little tiny comment link. And all the comment link does is create a comment edit box that should have been visible in the first place!

[dpb]

Maybe since there is a login, the interface should hide the "Answer" box to the original poster so only the "Comment" button is visible or other respondents' answer(s)?

[John D'Errico]

I had considered suggesting that. But there is another problem. Suppose someone, not the person who answered the question or the person who asked it, wants to add a comment? Suppose someone comes along and finds that question, and they want to add a comment? They too see only an edit box, that adds an answer. Often their answer is effectively no more than "Me too", but they still add that non-answer.

So all viewers need to see a comment as easy to add, not as a tiny button they need to find and click BEFORE they get an edit box.

[Oleg Komarov]

I agree. Taking SO as an example, they have a button that says "Answer you question" which if pressed shows the answer box.

Basically, the "comment this question" should be a bit bigger and the answer box for the OP should be hidden at first.

[dpb]

So maybe the interface should only have the two buttons "Comment" and "Answer" the question initially, no edit box at all. Then you've at least got a 50:50 shot... :) And, then, the users with sufficient privileges to be able to do the move/paste/delete operation now could be presented with a new option button to simply "CONVERT" one to the other w/o the cut'n paste operation.

Switching horses in midstream a little, while this is a problem, the interface issue that bugs me the most is the "CODE" button/usage of a a standard word-wrap edit box as the default. I'd guess 90% of questions posed have ill-formed code owing to that that is if not totally illegible, at least marginally legible. I know I and a number of other regulars reformat but that, too, is a terrible waste of time/effort to have to do so simply to be able to read a question.

[John D'Errico]

Yes. The "code" button is another terrible part of the interface.

[dpb]

Is there any part we agree isn't (terrible, that is)??? :)

I strongly suggested should have stayed with a newsreader-like if not actual nntp text format but with closed hosting if that was the intent back then...

[per isakson]

... and attachments are missing because attaching a file requires two steps

• Choose file and
• Attach File

and new users sometimes miss the second step (I guess)

[Adam]

I think not allowing people to answer their own questions would be slightly annoying, although admittedly only in a handful of cases.

There are occasions (I had one myself on the last question I asked) where you post a question and then quite naturally you carry on trying to solve it and come up with an answer before someone responds.

In my case I felt the question was still of value for if anyone else had a similar problem and did a search (unlikely I admit...) so I provided an answer to my own question rather than just delete the question.

I don't know what to suggest really though for the Comments/Answers problem. It is extremely frustrating, in common with many other things that new posters do (like the code blocks and putting as much of the question as can fit in the 'Question' box rather than a summary).

I guess I'm in the unhelpful camp of "I don't know how to design an intuitive interface, but I know a bad one when I see it"!

[Adam]

On a similar note the lack of anything in posting a question that encourages, preferably forces, a user to give their Matlab version (from a predefined list rather than an ad hoc string) is an annoying omission.

Some Matlab upgrades come with significant changes, e.g. the graphics changes in R2014b so knowing if a user is working with graphics objects that use . notation or still working with the old numeric handles is useful. Having to provide answers that use the old get(...) and set(...) notation just to be sure means pickup on the . notation by people asking questions here will likely be a lot slower.

[John D'Errico]

I've never suggested not letting people answer their own questions, nor have I seen that suggested.

It has been suggested that in order for someone to answer their own question, either they should have to click on a button that says, ADD AN ANSWER, or there should be an direct edit box for comments.

By having an edit box only for an answer, but NOT for a comment, it virtually ensures that people will make a mistake.

[Adam]

dpb's first comment seemed to make a suggestion that users cannot answer their own posts.

[John D'Errico]

Sorry. Many responses, I missed that from dpb. I agree that it should be possible to answer your own question. But if it is too easy to do so, then it tends to make any question into a newsgroup like thread, with a sequence of answers added every time someone wants to simply make a comment. That is what is encouraged by the current interface.

Logically, it seems that an answer should actively attempt to answer a question posed, else why would you offer the ability to attach comments that are distinct from answers? I think the problem I've identified comes about because it has been made too easy to add an answer, while making the person actively look for a place to add a comment, something that takes a second step.

I think it comes down to which of two fundamental paradigms Answers should tend towards. There is the newsgroup paradigm, where EVERY response is appended to a sequential flow of responses, the latest response appearing at the end of a linear chain. Since this linear chain can run into many pages of responses, it can become tiring to find the wheat from the chaff. I've seen many forum threads that run into hundred of pages of responses, most of which are of no serious value.

The alternative paradigm, that I thought Answers was built to follow is for an answer to attempt to answer the question at hand. Hey - I have something important to add to this conversation, something that solves your problem. Comments are something that should be easy to attach, perhaps to ask for clarification, even potentially to point out a limitation of an answer.

By the way, one class of "comment" that happens far too often (and often as an answer) is "Did you ever resolve this question?" This from a person who is looking to solve some problem, so they search answers, and find some question that seems related. So they add an answer to a 3 year old question, that merely says "Me too! I also have this problem." Of course, the person who originally asked the question is long gone, and will not see or respond to the "me too" followup. (Should there be a timed dormancy flag to add an answer? Where if a person responds/comments to a question that is too long dormant, then a box pops up that points out the timing issue?)

Finally, I've seen some forums that attempt to use the linear chain paradigm, but then sort the responses based on how many people like that response. Value sorting is an option, but it then distorts the time sequence of comments, so a response now often appears out of context. So we might have a sequence of comments A,B,C,D, which flow as a conversation. But if people attach lots of "likes" to comment C, then it gets sorted out of sequence from the rest and completely distorts the conversation. So while value sorting can work, it only makes sense if some relativity is retained. So any comment can have a parent comment, as well as children.

[Adam]

Even reversing the current format of big box for an answer, little hyperlink for a comment would be a big step forward. In general people giving answers know their way around the Matlab Answers forum and would still easily locate the answer link whereas those adding inappropriate answers that should be comments are mostly newer users who, as you say, find the most intuitive place to add their response and do it.

I never use the newsgroup, but I'm not really a fan of that approach to the problem in general.

[John D'Errico]

Yes, once people know their way around, there is never an issue. At least make it so for the first answer posted by an individual, an intercepting box pops up first that explains the difference between a comment and an answer, and asks them if they really intended to answer the question or if this should have been a comment. And if it is made at least as easy to comment as it is to answer, then they will get the idea.

[dpb]

My thought initially was the respondent can edit the question to announce he's got an answer or simply delete it rather than adding an answer. This is a little draconian, yes, but often the query really is rather low-level so didn't figure it was too much of a hit in actual content.

But, I did mollify the position to essentially agree w/ John that make the button choice the interface altho I think it's still highly likely whichever is the first one in the pecking order will get the majority of hits for the uninitiated unless a really major redesign were to take place (and no, I don't have a better one to suggest at the moment; I "don't do windows" (meaning I never have built gui interfaces so have no expertise whatsoever in the area)).

I still think there's a high probability that simply the facility to fixup these mixed up entries automagically given to the folks who can edit/delete questions now would likely lead to sufficient cleanup that the issue might become moot since it wouldn't be such a major hassle to do.

My primary reason for the newsgroup, Adam, is simply one of the much lower user overhead in reading responding via a newsreader than that of the web interface means I can do much more much quicker. And, I don't see that Answers has reduced the number of "eternal september" queries by becoming the de facto FAQ simply for the reasons outlined above.

I do like the facility to add attachments and to host screenshots or figures as it does sometimes clarify what the poster's real intent is and it is a more convenient way to get a dataset of size. (Of course, while most newbies don't "get" it, it's also true that a sample size of 5 is generally just as informative as 5000.) But there are newsreaders that can handle attachments (I use Thunderbird as one) but I can scan multiple newsgroups and respond to both cs-sm and comp.lang.fortran (as a couple of technical newsgroups I follow) in the time it takes to do perhaps half to maybe two-thirds the number here but I also have another 4 or 5 nontechnical groups plus OpenWatcom that I monitor. The newsreader lets me have those all in one place at a much less intensive user interface.

Answer 1

[Sean de Wolski]

I agree with a fair amount of what has been said. Here's a "constructive" way for y'all to help.

Randy is the main developer for Answers, but he has a User Experience person who is looking for new ways to enhance the page. I highly recommend asking them to do usability studies (you can email Randy or me, I'll forward it along). In theses studies, you can not only provide real feedback on upcoming features, but more importantly can rant about whatever you want!

[dpb]

Thanks, that is a useful piece of info...my complaint w/ the forum in large part is that it's more of a "high maintenance" form to read/respond to than a newsgroup so I'm tending to back off with other demands for time...the enhancements of attachments/formatting/editing just don't make up for the klunkiness and I don't think the idea of it being a searchable database that gets queries answered by users finding their problem already solved works -- not that the answer isn't likely somewhere already but that virtually nobody will search a priori and don't know how second and lastly wouldn't recognize it is the answer even if 1) and 2) weren't obstacles because it isn't exactly couched as their particular problem.

[John D'Errico]

I've participated in a couple of those studies before. It is not at all hard to do, and I hope I was able to provide some useful feedback.

Answer 2

[Jan]

There are no explanations on the page for posting a question. Neither the intention of the "Question" line is mentioned, nor the method to format code, the idea of the question/answer/comment fields.

The small "? help" button opens an identical page on a right-click, on a left-click just the mark-up is explained, but as we see in the frequently missing formatting, this list has no high impact factor.

There are smart ideas behind the interface, but they are not intuitively clear. Then some instructions are required.

[dpb]

Since folks generally don't read Help first as a rule, IMO the defaults should be such that what one would expect an uninitiated user to do should lead to expected results. For a coding forum, for the default edit/entry tool to be one with word-wrap "on" and not honor spacing and line breaks is simply the wrong basic starting point.

If that weren't the case and the edit box were to leave stuff as the user input it, I'd guess about 75% of the code posted would, while not necessarily be "ideally" formatted, would at least be legible and not require reformatting to even read well enough to try to decipher the underlying question raised.

I've not kept statistics, but quite a high number of questions I've edit'ed for the purpose simply need the two spaces inserted at the beginning of the first code line. If it weren't for it then by default reverting to wrapping after a blank line, a very high fraction would then be ok; as is, often have to either remove a bunch of blank lines or introduce the double-blanks for a number of paragraphs or the like to get clean copy.

That change alone (to turn off word wrap) would solve a high percentage of the formatting issues. It won't fix those who put the whole query in the subject line (limiting it to 30 characters (say) or less would, however, stop most of that.

If high-points users could simply convert a COMMENT to an ANSWER or the reverse with a button, could clean up that quickly enough that likely a number of the regulars would do so.

Adding another nuisance issue, when anybody does an EDIT on an Answer or Comment, the link to that (often the last) entry is broken. While get there from the Question link, it's inconvenient.

Answer 3

[Image Analyst]

One other flaw I've noticed is that the count of displayed answers is not correct. It will say 1-50 displayed, yet there is really only 15-20 answers displayed. Makes you wonder whatever happened to the other 30-35 answers . Were they perhaps spams that got removed and put into the spam quarantine yet are still being counted on the home screen as being displayed? Or is the count just wrong for some other reason?