A lot of energy is spent managing poor quality questions.
Poor questions create an enormous tax on SE sites:

  • Good users spend an awful lot of time flagging, reviewing, commenting, clarifying, and editing poorly asked questions. That time that could be much better spent answering or asking questions.
  • Lousy questions choke up search results, occupy space in user feeds, and add complexity and duplicates to the site.

I think a better UX would help a lot by reducing the poor questions at their source.

Here is the current SO question form:


The highlighted (yellow) panel focuses on showing users how to format questions correctly. I think that panel would be better used to show users how to ask good questions. I would much rather have a better quality question than a better formatted question.

For example, most users want good answers, so it would help to draw a direct connection between: (a) good answers and good questions; and (b) poor answers and poor questions. This provides an explicit behavioral incentive for users to ask good questions.

Here's an example of a better UX approach to avoiding poor questions:

new form Larger image is here.

This design repurposes the panel towards question quality, articulates the behavioral cost/benefit for users, connects the panel visually with the text box, uses more saturated fill to draw attention, and reduces the number of bullet points to a more digestible 4 (versus 8 today). I'm not sure I'm a fan of the multiple links (e.g. it could be replaced by a single link to something like this).


I agree with this in principle. The problem is, people don't read (UXMyths). Advanced Common Sense has some nice diagrams of what people see online. I'm reposting a couple (apologies on the low quality, it isn't my fault).

We design for people to read the page -- people just skim looking for keywords

Users only actually see a bit of the page

This holds true for your suggestion as well. I edited your images to show (the squared off stuff, plus a couple lines) how an average user (such as me) would approach it. I might even be too generous with what people read.

The existing form:

I see textboxes, nice textboxes for writing in

Your new form isn't much better. :/


I just might notice the triangle, and then go see the "Want a good answer? Ask a good question" header. More than that? Fuhgeddaboudit.

So, if you actually want people to see this, you're out of luck. People will still just ignore it. Maybe one or two people will notice it and ask a better question. But the majority? Fuhgeddaboudit.

  • 4
    Not constructive or correct. Your argument could be used to show that all forms design is basically pointless. I do UX for a living and it's testably true that form cues and callouts work. If the visibility is too marginal, the the solution is to adapt the layout, not abandon it. As a reductio ad absurdum example, if the callout were bright yellow with black text and placed on the left, and the form fields had rgb(250,250,250) borders and font color, I guarantee the callout would be read. – tohster Mar 2 '15 at 19:36
  • It might be worth taking another page from that same book and do a little user testing to see if it might help. While I'm cynical enough to agree with the general sentiment, I'm hopeful enough to at least give it a shot. The only snag is how would you define "improvement" and what test would show "improvement"? – AnonJr Mar 3 '15 at 18:18
  • 1
    Totally agree that user testing is helpful here, as it is with most things UX. But we don't have to stab the darkness here: form design is a well researched field. Saying that "people don't read" is like saying that "people don't pay attention" so nobody should drive: it's useless cynicism and results in very wrong conclusions. Here's just one intro-level article on form design that includes notes on eye-tracking and prompts...there is tons more out there on good form design: static.lukew.com/webforms_lukew.pdf – tohster Mar 3 '15 at 21:18

You must log in to answer this question.

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