A Case Study of “Designed By Developers:” Stack Overflow

1. When I log into stack overflow, the status bar at the top usually contains information I want to know.

2. The use of negative space is better, but the question title and question body seem to be too separated from each other. The question body looks too much like an answer, and their isn't enough differentiation between the two. In the original design, the tabs and the line clearly separate from the answers.

3. He seems to have gotten rid of a lot of the statistical information about the page. I use this information to decide on the value, relevancy, and recency of a question. It's kind of like Github, one of the first things I do is check the network graph, and see how healthy a project is.

4. The votes are too small in his new design. This is a critical aspect of the site, and allows visitors to make snap value decisions. It should be large, it's at least as important as the entire body of the message.

In general, his design looks better and is less busy, but would kill the overall usability. Stack overflow is useful because of the information density, not in spite of it.

Let's take the bright orange banner at the top of the screen that appears only to new users.

Obviously, it's not just new users that see it. It's anyone who hasn't got a cookie on their browser indicating they haven't closed the ugly bright orange banner before, so it's hitting plenty of people who know this information already.

But let's ignore that, and ask whether it does the job: how useful is the information contained in the big ugly bright orange banner to new users? Not very, I suspect. For most users, the purpose of the site will be implicit - from context (because they did a search for their problem and were linked to it), from the design (it looks like a Q&A site), and from the content (questions and answers). If it's not implicit, then the exact same information is explicitly stated in a box on the right-hand-side of the site. No other popular modern site on the web does these sort of popups anymore, so I don't see why the creators of Stack Overflow thought they were a good idea. In addition, why does the big ugly annoying bright orange banner link to the FAQ? What pecentage of new users will ever want, need, or read the FAQ? Users don't read anything. If the idea is to encourage sign-ups, then create a sign-up link, not a link to something nobody will read.

Then there's the functionality of the big ugly annoying intrusive bright orange banner. It loads after the rest of the page loads, then shifts all the elements of the DOM down. Just in case you'd missed the bright orange banner that distracts you while using the site, it scrolls with you, displaying on all Stack Overflow pages until you're forced to go out of your way to move your mouse to close it (with their weird 'close' icon), when finally (if you're near the top) it shifts the DOM up again, just to ensure it can be maximally irritating.

I don't see how "you’re seeing the one-time new user screen layout" should be regarded as any sort of defence for the big ugly annoying intrusive distracting irritating bright orange banner. Do new users somehow deserve bad design and bad functionality?

In my eyes, Stack Overflow has firmly been a success despite its poor design. It's not as though Experts Exchange set a very high bar to beat usability-wise.

Most of the users who simply read content, are looking for an answer to a particular question. Chances are, they're coming from a Google search for a specific problem, and SO has the answer. Their actions and expectations are geared towards finding the answer as quickly as possible. I can't say for sure, but I think most users don't have much problem seeing the question or scrolling down to see the first answer. They also probably realize it is the "best" answers by the score next to it.

On the other hand, you have the regular users. Most regular users don't see the ads which bombard the newcomers, which already removes a lot of the ugliness from the screenshot he posted. Also, as a regular user myself, a lot of the other stuff he claims is "useless" is very important: the info bar at the top, the rep scores, the tags, all of it.

I'm not saying Stack Overflow is perfect. But to understand its design, you first have to understand the different users who come to it, and analyze whether the "off the street" user manages to find their answers or not, and whether the site is good for the "community users" (obviously it is, considering its success).

By the way, if anyone is interest, Jeff himself comment on the blog.

Jeff launched this thing by tuning it for what he wanted in a Q & A site. Since he is exactly the sort of early adopter they needed for the first one (for programmers) that worked OK. But now they are launching a fully baked, "Jeff-tuned" product to other populations and I think there will be problems. The problem with an overloaded page is that you've closed off your options - you've made it really hard to iterate. Because every piece of available space is filled with information it's very difficult to test the product with new users.

This happened partly because this is every programmer's instinct, and partly through Jeff's misunderstanding of Edward Tufte's concept of "information density". Somewhere in a Stack Overflow podcast Joel is telling Jeff that there is way too much stuff on the page and Jeff defends that as being an example of information density that Tufte would like. Not true - a vector field has information density: http://www.google.com/images?q=vector%20field When you can apply a single (simple?) rule to understand a lot of data, that's information density. By contrast, Stack Overflow is like any web page in that each element must be examined in turn to be understood. That's why you must eliminate elements ruthlessly, especially when you are starting out.

As the article's criticisms have some validity, I do wonder - does Stack Exchange need to revisit the information architecture of the original code if they want it to be the Q&A platform for the masses?

More seriously, the link above shows how early on Jeff Atwood got design input from the people who use the site. Jeff's comment on Dan's critique also show how they continue to work with the users on design, and they've hired a designer, Jin, who also commented.

The post is like a small version of the American Airlines design critique post from a while ago, complete with a polite explanation of why the design is the way it is.

Seriously? What a horrible way to tear apart someone else's hard work.

An actual case study of Stack Overflow would be that something designed by developers could be a huge objective success.

Also, his proposed improvement "hasn’t included all the functional elements yet" Anyone can make almost any interactive design appear to be superficially better as a static screenshot by taking things away. As far as making something that actually works and does the right thing, that takes more than Photoshop and arrogance.

It's sad to see this guy getting attention.

It's not that his version is bad but redesigns by outsiders necessarily introduce their own new flaws. They don't know what constraints drive a specific design, so there's little to learn from a mockup that fixes, say a white space issue, but did not consider if usability or conversion rate could be impacted.

I came to stackoverflow from search engine multiple times while searching for solutions to my problems. I was pleasantly surprised how fast I could find question that was asked and after confirming that the asker has similar problem to mine how fast I could browse through responses. Also I could get to know reservations other user had to responses (stated in comments to those responses). I shouldn't probably be surprised by my amazement because the thing that recurrently cropped up in my search result in old days was experts-exchange.

After multiple visits I finally got curious what is that site that helps me so often and I clicked link in the orange banner on the top that drew my attention. Up to this moment the only parts of the page that I've noticed were, questions, answers, comments and banner on the top.

After I registered and answered few questions I picked tags that interest me. They are visible on the right. Tags are for answerers not for askers. Tags visible to unregistered users are just cool unimportant bonus info similar to pageviews and they are placed in the bottom right when nobody even looks if he's not bored and exploring.

There are maybe some features that are not necessary, and some missing but what is important for me, no unnecessary feature ever got in my face while using this site.

As for the proposed redesign I think it's horrible.

You can barely see votes and up/downvote buttons that are very important for this site to work. If people can't find them and hit them less often more bad answers would be mixed with good ones.

User info is in prominent place and it is almost of no importance ~8k rep guy can give you bad answer. Votes on answer matter, not the rep of answerer. Also insane spacing around it immediately kills aesthetics of the whole design it might have possessed. I rarely see something that awful outside of works of beginner designers (and yes, also developers).

Putting main menu in the right top corner where are all the things that you don't usually care about until you want to search for something or log in is a bad idea. But I don't use this main menu all that much so probably it wouldn't hurt too much. ... Besides if you see input box in the top right corner what do you expect it to be? A spot where you can place your pizza order? Even if that box was empty as long as it looked as input box I'd have thought that this is serchbox and when I put some stuff there and hit enter site will be searched for occurrences of it.

Questions are just barely discernible from answers. Stackoverflow does this better but I'm not sure why. Maybe with spacing? Maybe by not discouraging you from reading it by putting it somewhere in the middle of gray background?

Only thing I think might be good in the redesign is exposing "Ask a Question". But I think SO does almost as good by surrounding this option with plenty of whitespace.

As a footnote:

Don't try to redesign how thing looks until you understand what it does.

"By developers for developers" sometimes works.

Obviously, this guy proves that it goes far and it applies to a website such as stackoverflow as well. When I came to stackoverflow 1 year ago I had an amazingly streamlined experience. But of course, I had previous experience from using the internet. I expected that

* the site search box to be in the upper right corner, and use the verb "search" or "find" * the login/register links to be in the same upper right corner * most publishing tools have a permalink to be found in relation to posts on the page * 'flag' is likely related to fighting spam, a problem most open forums have * 'tagging' is that which was made popular in this whole web 2.0 craze * "Ask Question" would mean something involving "asking" and "a question" * an up/down arrow, above and under a number would increment or decrement that number * it is beneficial to place a mouse pointer over items to see if there is clarifying mouse-over text, but with little fear I tend to push buttons to see what happens as well

Complaining about ads by guessing they're not effective, or that the careers link is just hidden ads also seem quite malicious. They got to earn money, and having read jeff's/joel's blog entries on the success of ads on stackoverflow I am very sure that they've been trying out different things to see what works.

It is difficult to criticize these remarks without sounding a like a wise-ass. He does have remarks about sensible things, such as the graphical element used for sorting answers, or the background colors of areas that have static and less than important information. But I guess this guy primary mistake was to frame his remarks as "these are things that need to be changed, because im a designer and user experience expert", rather than "these are things that would be interesting to perform A/B-tests on to see if they improve the experience". That is how it came across anyway.

Also, the author's choice of having the masthead of the blog grayed out does not seem consistent with his criticism of gray type in the design.

Being a good developer is not about typing code, it's about creating a great application. That means using all of the tools at your disposal... code, visual design, workflow structure, analytics, etc.

In my experience, traditional designers often do very poor work when it comes to interactive applications. As a developer, it's my responsibility to tell them it's crap, why, and present a better solution.

Obviously being butt ugly didn't hurt them that much.

I find the tone of this article a little annoying in that it plays to the businessman / designer stereotype of developers not knowing anything about design, and only being focused on functionality - to a fault.

His idea of paying for answers (near) instantly would be handy though sometimes.

Why do you need a good navigation if people just hit the website to lookup the answer? If they want to become active members they'll learn it!

The font and colors choice are quite good for reading. Reading for long hours and long periods with this theme without getting tired or annoyed, I should say it was really well designed.

New users get answers easier than on any other site. Users who understand the value of the site will spend time to learn become valuable community members. Lowering the barrier to entry would lower the quality of the content.

Main point being I focus exclusively on the center content portion. Ads and BS have trained me to completely ignore sidebars and mostly ignore headers(some sites the header/center content is not clearly seperted). I never look at banner/menu until unless I want to explore site.

[1] http://img713.imageshack.us/img713/3498/soregistereduserscre... [2] http://danzambonini.com/wp-content/uploads/2010/07/stack_ove...

I understand that these sites need to generate revenue and advertisements are one way of doing that, but given the eye-gouging (flash-gif-html5) ads designed by most advertisers my sympathy lies with the readers who visit these sites for reading content.

Everything else isn't a rocket science - launch fast/early and evolve! ^_^

There are tons of similar ideas in non-IT fileds (travel guidelines or a-la AirBnB for hotels/trekking agencies in third world - the first that comes to mind). But who will pay my bills?