When should we make a conscious decision to retain poor UI, even if temporarily? Not because of budget or time constraints but because of what I call the QWERTY paradox.

In short, the QWERTY paradox is that users have a better experience with a poor interface than with any of its contenders because it's been so widely learned.

QWERTY Paradox: The letters on a QWERTY keyboard were originally arranged to prevent typewriter typepresses from jamming. Of course, this is a not an issue today and the arrangement of the letters are actually rather inefficient.

There are several more efficient layouts for keyboards which have been around for a while such as the Dvorak keyboard. However, a good computer designer with a passion for usability ought to ship a keyboard with a QWERTY rather than a Dvorak keyboard.

The former has been used so often, users are much more efficient with the convention of QWERTY than with the unfamiliarity with Dvorak.

So my question is, On what criteria should we opt to retain bad UI to keep our UX pleasant (paradoxically)?


I think the point of the QWERTY paradox may have been lost. Essentially it's this:

The gain from an improvement can be significantly less than the friction from migrating to it.

  • 4
    Is it possible to offer the choice? People are resistant to change, but intrigued by novelty. If you offer them the chance to try the new UI (or force them, but offer the chance to go back to the old) many will willingly convert if it really is better.
    – aslum
    Commented Apr 17, 2012 at 18:00
  • DVORAK isn't necessarily better than QWERTY. There are 3 ways to make an efficient keyboard, and QWERTY focuses on one of them while doing okay-ish on the other two, while DVORAK optimizes the other two while completely abandoning the first one. (They were something like: how often the hand used alternates, adjacency of commonly used keys, and I've forgotten the third) Unfortunately, I'm unable to find that article at the moment...
    – Izkata
    Commented Apr 17, 2012 at 18:02
  • 5
    I can think of some evil reasons for bad UI.
    – zzzzBov
    Commented Apr 17, 2012 at 18:19
  • 2
    Also don't forget that sometimes Ugly interfaces can be a good UX. But it does seem you mean bad UI in the functional rather than aesthetic sense
    – Zelda
    Commented Apr 17, 2012 at 18:48
  • 2
    @Izkata as far as I know, that's not true. See mkweb.bcgsc.ca/carpalx/?popular_alternatives for example. QWERTY is pretty bad in every aspect. That letters on a QWERTY keyboard were originally arranged to prevent typewriter typepresses from jamming is an urban legend, however. QWERTY was "optimized" so that all characters of the word TYPEWRITER were put in the top row :) The purpose of this was to make it easy for first salesmen to demonstrate how the thing works. I've leave for you to judge whether it's likely to make QWERTY an overall good keyboard layout Commented Apr 18, 2012 at 9:16

11 Answers 11


Only when a change would be absolutely earth shattering. Changing the standard keyboard layout would mean hundreds of millions of people can't type any more. It's a change that affects the majority of keyboard users the world over at this point.

It might seem like this is a good reason to not change things when the change is disruptive, but it's really not. Most changes people can get used to over minutes or days. The problem with keyboards is that these are physical devices that can not feasibly be replaced. If you change keyboards, hundreds of millions of the old ones are still out there. There's no reason to change. You can't force the change.

In the digital space, change is almost always possible; Windows Vista made a great deal of people freak out, and the Word 2007 Ribbon confuses new users to this day, but you can adapt to them (especially due to the significant UI improvements in Windows 7 and Word 2010).

If you can force a change, people will get over it. You better have a good reason to force the change, as there will be short-term consequences, but if research really does indicates performance improvements, the change should be made. Allegedly Microsoft did find significant evidence supporting the Ribbon, hence their change.

The reason you really need to force a change is the longer a problem remains, the bigger a problem it becomes. Changing keyboards would have been easier in the 70s. If computers used different keyboards than typewriters it would have been a little annoying at first, but keyboard penetration was small. Now the level of inertia is almost completely insurmountable.

Change Fast when research supports it. The digital space has a lot more room for change than the physical one for many reasons, which is why you need to be sure physical things work before you distribute them.

  • 1
    Excellent points. There are certainly differences between a widespread standard like the QWERTY keyboard and something like a company's website, but this is just used as a sort of flagship example. I like your advice to "Change Fast when research supports it."
    – Adam Grant
    Commented Apr 17, 2012 at 14:50
  • 4
    "Most changes people can get used to over minutes or days." This may be true, but it assumes that they are willing. If it's very easy to switch my competitor's product for example, that might be easier than minutes - days of learning.
    – Adam Grant
    Commented Apr 17, 2012 at 14:57
  • I dont like the analogy as Ben points out. This is not an example of bad interface design per se
    – colmcq
    Commented Apr 17, 2012 at 15:00
  • 1
    @ajkochanowicz that's a complex issue of it's own. If you're an operating system/big player, they're probably willing. If you're small, you might be more interested in making things better for NEW users than retaining your old users. Either way, I don't think keeping the bad UI is a good strategic decision. I have to assume this question is about strategic changes rather than tactical. Change is always bad for the first 5 minutes.
    – Zelda
    Commented Apr 17, 2012 at 15:04
  • 2
    +1 "If you can force a change, people will get over it [...] there will he short-term consequences." A perfect example of this every single iteration of a Facebook UI redesign: everyone complains and threatens angrily to abandon Facebook, then everyone forgets what the old site was like a week later. Surely, replace Timeline in 9 months with Facebook circa 2008 and everyone will hate that, too.
    – msanford
    Commented Apr 18, 2012 at 20:26

I worked for years in e-commerce development and one of the things I learned the hard way multiple times was simple: No matter how terrible UX is, and no matter how much of an improvement your new design is: If it has an established user base, you can't completely change it overnight. Imagine if when you went out to your car in the morning everything was moved around, the door handle was in a new spot, you couldn't find the ignition, the gas gauge wasn't where you remember it, the seat belt functioned in a new way... you would hate it, because it would slow you down at first and all you really want to do is just drive to work... Even if in the end it would have been a better layout, you would have wanted your old car back immediately because you just needed to get things done.

Every time I've ever been on an e-commerce project that completely changed the look and feel, even for a newer, extremely well-planned and well-executed UX: The sales absolutely tanked for months after it rolled out. Over time, they would always recover and generally come out better than they were in the end, but initially, they would drop, sometimes more than 50%. There are a few reasons for this, mostly speculative but: People would come to the site and think it was some other site or that the company had change somehow (ownership, product line, etc). Or people would come to the site and not be able to find things they used to be able to find and they would ditch.

To mitigate this, we learned to spread out the roll out of complete UX overhauls into phases, where we would introduce new elements and/or move things around bit by bit until the ultimate goal was reached. For example: the location of the shopping cart didn't change for a while. First we changed the color scheme to something that suited our needs a little better, then once that was established, we were able to move the cart to it's new home and people found it okay because it's look didn't change, just it's location.

  • Sometimes, drastic changes are made successfully: Samoa switches to driving on left Commented Apr 18, 2012 at 11:02
  • As a side note - it's important to measure incremental changes in the UI, preferably with A/B. Full redesigns result in number shifts which cannot be attributed to a specific element. For example - if you overhauled a shopping cart which resulted in a sales increase in for "one-time shoppers", but a decrease in sales for "return shoppers" with a net effect of no overall sales increase - what would you do? Roll out the product and risk losing higher value customers for the sake of a "better UI"? Incremental and measured changes would teach you which components work for both customer types. Commented Apr 18, 2012 at 13:05
  • 3
    IMO, we can learn something from how Google does it all the time. For instance, "The old Google Groups will be going away soon. Switch to the new Google Groups." Tell folks that the change is imminent, and you better get ready. It's better than changing everything overnight. I am no UX expert (nor from this field), but a normal user, and this is my simple understanding. :) @blesh you nailed it, by the way.
    – its_me
    Commented Apr 19, 2012 at 6:31
  • @geekpanth3r, you're very right. That is one technique we used to introduce people to a new site layout. But even then, on our most successful change-overs, we didn't do a complete change. It was still in stages.
    – blesh
    Commented Apr 19, 2012 at 13:21
  • 1
    @blesh hmm... makes sense. I simply don't understand why no one documents the changes. It would actually make the changeover a whole lot smoother for the users. Not even Google does this all the time. I am loving the questions on this site. Great insights.
    – its_me
    Commented Apr 19, 2012 at 13:23

You can use a bad user interface to signal attractiveness/unattractiveness to a subset of the audience that's likely to visit the page.

For example, the Hacker News website is fairly ugly, and Paul Graham has commented this is by design. The goal is to attract people that want to be there for the content and don't care about how pretty/cool it looks to be participating on that site.

  • 4
    It may be ugly, it's simplistic, but it doesn't mean it's "bad". It's clear and intuitive. Commented Apr 18, 2012 at 9:39
  • 2
    @Morawski it's kinda bad, hard to read, very tiny and hard to interact with.
    – Zelda
    Commented Apr 18, 2012 at 19:57
  • Agreed, @Ben. One can still be plain and unpretty while still be comfortable, readable, and easily usable.
    – Adam Grant
    Commented Apr 20, 2012 at 18:34
  • I think aesthetics might be subject to individual preferences, so to the intended audience the design might not look that bad (although you'd probably get a different opinion from a graphic designer). In terms of functionality though, it is hard to hide bad usability with nice visual designs - but it might be easier for people to accept it.
    – Michael Lai
    Commented Jan 6, 2015 at 22:20

I think it's most important to consider the return on investment you would get from improving the interface. Assuming this product is being actively developed, it might make sense to make some minor improvements/shortcuts for advanced users (who are ostensibly not too dissatisfied with the UI) but leave drastic improvements for later versions. Users would be able to slowly roll out the new version and train/assist as necessary.

Take what you've learned about the interface (i.e. why it doesn't work) and use that to make the next UI excellent.

  • Yes, this is more along the lines I'm looking for.
    – Adam Grant
    Commented Apr 17, 2012 at 17:53

I'm going to go a little sideways on my response...

A good reason to use a bad UI (from a business perspective) is to push a user/customer into the optimal channel for the transaction.

For example: You may allow a customer to cancel a service online. You may work in web features which try to save the business, but they will never be as effective to save the customer as directing them to a phone rep. Providing a poor experience online might convince them to call to cancel providing a better opportunity to save the relationship.

That's a business reason for bad UI anyway.

  • 3
    that's more of a dark pattern than a good ux reason however
    – Zelda
    Commented Apr 17, 2012 at 17:56
  • 2
    I don't disagree and I did qualify my response as a "business perspective" on a problem, but I'm not sure you can say it's a poor UX reason. UX is the compromise between business goals and user expectations. If the web app can recognize "save-able" customers and direct them to a channel that meets their needs (resolves the issue)...I can't see this is a bad thing. Commented Apr 17, 2012 at 18:10
  • "UX is the compromise between business goals and user expectations" UX work, perhaps.
    – Adam Grant
    Commented Dec 20, 2013 at 16:00

One reason to keep ugly interface design is to set expectations properly early in a project. When showing off things to individuals or clients who are not a part of the technical team, you want to set realistic expectations as to the progress of the project.

I have found that if the interface works reasonably well and can interact in ways to cover most workflows, then people will think it is much further along than it actually is. This is very common in rapid prototyping with web technologies.

If you keep the flash and styling to a minimum (even uglifying something sometimes), then people won't jump to conclusions so quickly about the product being nearly finished.


I've said this before somewhere else - but will repeat it.

If you use a familiar UI users will attempt to guess their way through how it works. This can often lead to sub-optimal patterns of usage.

If users don't use the software much this doesn't matter.

However if the users job is to use one piece of software day in day out - they really need to be using it in the most optimal fashion. So they really need to read the manual / go on the training course - rather than just thinking they can guess there way through it. So it helps to make it look as unguessable as possible.

A second factor in this situation is that users will build up lots of expertise on the system so they need as need as little interface to deal with as possible.

So taken to the extreme the ideal interface for the software may be purely a command line:

...ie c:\


using qwerty as an analogy for bad UI is going to influence answers, since changing the keyboard, for most users, would render a computer useless. for a computer manufacturer this would be financial suicide.

however, if we are not influenced too much by your analogy, and focus solely on the question, not all changes require this kind of compromise from the user. probably, more of them are not so catastrophic. for these kinds of less catastrophic innovations, i would say they should be welcome, given they follow solid, proven principles.

yes, it's a fact, even an innovation that clearly makes the user's life easier, may not work at all. on the opposite side of the coin, if enough people do the wrong thing, it ends up being the right thing. a huge example is how we always look to the top left of the screen... that's a prime piece of real estate now, but would have been pretty confusing in ancient china, when poetry read from right to left, top to bottom.

as ux designers we should be welcoming to innovation and evolution of the ux. if it's not us who create the interfaces of tomorrow, then who will? programmers?

at the very least, a ux designer should be sure that their application will be competitive with what is currently out there, whether with competitors, or with other sites that have a similar flow. once that is achieved, then i'd say this kind of innovation should be explored.

i wouldn't always agree @ben above who suggests that the change should be earth shattering. a lot of innovations are subtle, and are cumulative in the way they allow the user to more freely and subconsciously operate the application. if you look at the web, everything that is successful has taken some chance of innovation. even if that innovation is copying something that is huge and just tweaking it a bit... that's even more ballsy than pure innovation in a sense.

look around on the web, it is evolving every day, new innovations are introduced daily. it's like Darwin's natural selection in the virtual world. some things work, and some things don't. those that don't work will result in death, those that work will result in elevation of the species, and those that just copy innovation are probably generally safe, but no guaranties. so perhaps it depends on your personality type, or perhaps you are the kind of person that is more successful copying others...

there may even be a stigma with ux designers that create appealing looking interfaces but lack in ease of use. i'd even be willing to bet that most ux designers are guilty of this. so perhaps they should take @ben's advice above, and not move forward with these ideas unless they can confirm their earth-shattering-ness.

  • "using qwerty as an analogy for bad UI is going to influence answers, since changing the keyboard, for most users, would render a computer useless. for a computer manufacturer this would be financial suicide." That's my point. The QWERTY keyboard is a great example for what I'm describing for exactly the reasons you mention.
    – Adam Grant
    Commented Feb 4, 2013 at 21:49

"QWERTY" keyboard is the situation when "bad idea become a standard". When somebody invents something - he does not feel responsibility for maybe creating new rules. No matter if these rules will be good or bad - they could become a standard. My opinion: if the design (rule, standard etc.) seems to be bad - it is better to change it right now.

  • This kind of restates my question. The QWERTY keyboard is a bad design gone standard. So if I make computers, it's best to make them with non-qwerty keyboards?
    – Adam Grant
    Commented May 18, 2012 at 20:06
  • @ajkochanowicz You may use Dvorak solution en.wikipedia.org/wiki/Dvorak_Simplified_Keyboard Qoute: Dvorak proponents claim the Dvorak layout uses less finger motion, increases typing rate, and reduces errors compared to the standard QWERTY keyboard.
    – webvitaly
    Commented May 19, 2012 at 22:57
  • I think you may need to read the original question posted.
    – Adam Grant
    Commented May 21, 2012 at 20:47
  • @ajkochanowicz you wrote "The former has been used so often, users are much more efficient with the convention of QWERTY than with the unfamiliarity with Dvorak." - it could take about few days to learn and get used to new keyboard layout, but the efficiency will increase. So IMHO it is better to use better UI rather than bad UI. QWERTY was created for have TYPEWRITER word on the top row for faster typing this word and for better selling typing machines. It is now became obsolete and it is better to choose new better layout standard.
    – webvitaly
    Commented May 22, 2012 at 5:56
  • If you became the Apple CEO overnight, would you make MacBooks use Dvorak Keyboards by default? Customers would come into Apple stores and all the computers on display had Dvorak keyboards?
    – Adam Grant
    Commented Dec 20, 2013 at 16:03

Inappropriateness & obsolescence might also be a factor A classic example - often kicked around UX circles - is the continued usage of 'the envelope icon' for email. Or 'the floppy disk icon' for [Save]

The argument is that many younger generation folks have used neither a paper envelope nor a floppy disk ... ever.

At what point is it advisable to replace these historical-physical-object-oriented icons with something more timely? or flexible? or conceptual?

The comprehension issues include:

  • Is it recognizable? (i.e. rendering)
  • Is it global (i.e. not culturally-loaded)?
  • Is it understandable? (i.e. do you get the meaning)
  • Does it have "legs"? (i.e. Will useful meaning persist over time?)

The trade-offs include:

  • Meaning: Object-based vs conceptual
  • Change: Accepted usage vs Learning curve

The consensus in the UX community re "whether an admittedly outdated but well-understood-icon should be replaced" was ... definitively mixed. Net/net: It depends. Testing. Research. Budget. etc.


Thanks everyone for your answers. I wanted to post another thought on this that is tangentially related.

I've noticed a paradox when creating prototypes, sometimes napkin drawings serve better to convey ideas than high fidelity, esp. functional prototypes.

When UI drawings are crude, it forces stakeholders to think about the ideas conceptually rather than to get in the weeds on the details of the design elements. It's an unspoken way of conveying "There would be something here that serves this purpose" or "This comes before that," etc. without having to add "But don't worry about which color this is or how thick the border is for now."

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.