home | content | archives
(posted by Pierce on January 20, 2005 at 01:02 AM)

AskTog.com has released a list of the "120 Most Unpopular Design Bugs", an extension of a December article of Bruce Tognazzini's top ten most persistent design bugs, at http://www.asktog.com/Bughouse/index.html.

A lot of good points, but a few caught my eye...

"Stupidly sticky slashes", under the Word Processors category, doesn't seem important enough to be in the top 120, but maybe that's just my opinion. The only people who are affected are those who use extended/series/of/slash/separated/words as a way of lazily creating lists of things without creating the proper description and outlining normally required by the language grammar. I'm guilty of this as well, but I honestly don't think it's worth making a whole preferences option toggling breaking/nonbreaking slashes (the proposed solution) since AFAIK they *aren't* supposed to be breaking, at least in English. Having a mostly-empty line before long slash-delimited lists seems like a small price to pay for not having yet another minute detail to think about.

Update: after publishing this, I thought of one example where breaking slashes would please me, and that's URLs in text documents. URLs get obscenely long at times, so they become very hard to inline into your paragraphs without the ugly aforementioned space. The only exception I can think of to my position above would be smart recognition and line breaking of URLs, since a URL already breaks the grammar rules of every human language I know of.

"Rollovers", under Websites and Browsers, also seems like a nonbug. I mean, there's no question that certain rollover images still reflect the rotating-envelope-animated-gif aesthetic era of the web, but they're increasingly in the minority. When I move my mouse over the "refresh" icon in Mozilla (modern theme), it gets a subtle hue. This tells me a few things... one, that my browser is still responding to user input, which is an important detail since I run a lot of computationally heavy tasks. If it doesn't happen, I know that either the program has crashed or it's trying to swap itself back into memory. Two, and more importantly, it gives me a tangible indication that *this* is the action that will occur if I click right now, not "forward" or "stop". I get the same benefits from actual web pages that use this functionality.

I don't find this bug to be a violation of UI principles, since it's (typically) not an *ugly* rollover button, and it's not distracting me away from something I'm trying to use. On the contrary, it is pulling my attention towards the thing I *am* trying to use, so I think it's beneficial.

Then again, the author claims to not use Office, IE, or Netscape, or presumably any other application with rollovers... so nothing developed after about 1995. I'm perfectly happy to let this guy wallow in his smugly superior lynx-and-emacs paradise, but I'm hard-pressed to accept his criticism of the mainstream world's progress.

"Ha! Ha! We donít ship there", also under Websites and Browsers, has an interesting footnote that occurred to me when I was reading it. Annoying users in geographic locations you don't ship to by making them fill out the entire order before the application fails doesn't seem to have a downside, since the people annoyed by it can't be your customers anyway. The author dismisses that concern with an anecdote about being on a business trip outside the delivery region, and a plan to hold the grudge when he got back within the area. But how often does that really happen? All the billing systems I've observed empirically do their order processing in one big check-shipping-and-billing-and-inventory-at-once step. It's not the best architectural design, but it allows easy system consolidation for companies that don't really need best-of-the-best interfaces. The engineering time required to rearchitect these systems into multiple distinct steps is almost certainly not worth the cost of the two guys who get annoyed and bitter the one time they can't get an order *that they wouldn't have been able to get anyway*.

"Writeable CD/DVD's aren't like other disks", under Bugs Across Many Operating Systems, shouldn't be a bug because it's true. Writeable discs *aren't* like other discs. It takes a long time to write to them, you can never truly delete what you've written on the -R versions (though you can tell the OS to ignore things that are there), and an unfortunate consequence of that limitation is that writing 600Mb onto an 800Mb disc, then deleting it as you would on a hard drive, would leave you with 200 free megabytes rather than the expected 800. I agree that operating systems should include reasonable interfaces to disc media so you don't have to use Roxio's buggy software, but those interfaces fundamentally can't be the same as hard drives and network mounts.

"Keyboard / Phone keypad layout bug", under Hardware & Drivers, would seem to be a hard one to argue against. UI consistency has been the mantra repeated by legions of HCI specialists. But I can think of a very good reason why phones keypads and keyboard numeric pads are reversed. All else being equal, the obvious way to orient numbers is as if the user would be reading them and counting up from zero (never mind that the zero is at the bottom of the keypad, something I never fully understood). When you're holding a cell phone, it's just about as easy to hit the one as it is to hit the eight, so that logic reigns. But keyboard numeric pads are used to enter many, many numbers into spreadsheets, data processing applications, and others. I don't have evidence to prove this, but I'm betting that the most frequently used numbers are zero and one. So it makes sense to have those numbers be the most accessible to the user. This means they should be closest to the place where the user is resting their wrist; the bottom of the keyboard*. That way, the user doesn't have to strain by "floating" their wrist over four and seven every time they want to hit one.

* Note: I know that the real most "comfortable" place to put the frequently-typed keys is in the middle row. But we have to eliminate that as an option because I think we can all agree that "789"/"0123"/"456" would be a sadistic keypad layout.

Honestly though, how many people are impacted by this difference in practice? Who types *so much* on phone keypads *and* computer numeric keypads that they are simply unable to fight their instincts when they switch between the two? Consistency is one thing, but if you have to supress comfort and functionality to get it then I think you've got misplaced priorities.

Under Programming & Command Lines, "Text-based programming languages" as a "bug" is taking the word to its most generic possible usage. It ignores that textual programming languages are unparalleled by any other software development paradigm. The author makes a coy reference to violating an NDA if he describes a solution more clearly, but I think we're still a ways away from any widely-accepted alternative that works as well or better than good ol' C, Java, Perl, whatever.

The generically named "Programming Languages" bug decries languages that syntactically permit poor formatting. In my opinion, lack of readability in poorly-formatted programs isn't really a bug any more than it's a bug that Microsoft Word allows you to write poorly or have strange outline numbering. The bug is classified in the "solving the wrong problem" category, but that's B.S. The problem that programming languages try to solve is the need for humans to tell a computer what to do. Programming languages just aren't meant to be used for describing a logical process to another human, otherwise every operating system would have to come with a built-in pseudocode compiler.

Don't get me wrong, I wish the code *I* work on was always clearly indented and variables always had proper names. But having style guidelines *is* the proper way to achieve that, not by having the compiler pretend it can't understand things when it very well can.

Now, don't let the above commentary fool you into thinking that I regard the bug list as an inaccurate waste of bits. The vast majority of the bugs were spot-on, and I'd love to see them addressed. I just wanted to vent about some of what I considered to be people pushing their own idiosyncratic preferences as absolute HCI truth.

Oh, and as a message to Mr. Tognazzini: in my humble opinion and with all due respect, no self-respecting computing professional should use Microsoft Office to generate HTML, *especially* on a multi-page site. "Smart" quotes became a capital O with a circumflex and a single quote, and (bizarrely) any instance of "Fi", such as in "First", became the Icelandic capital letter THORN. And when I looked at the HTML source to try to find unnamed anchor links, my brain developed a terminal hemorrhage from all the autogen<span class="msonormal">erated office for</span>matting littering the place. Just... just no.


The official reason why the phones have the number keys reversed is to mimic somewhat the ordering on the rotary dial that preceded them.

Posted by assen on January 20, 2005 at 03:51 AM

I guess I don't understand why that would be a mimic... rotary keys were in a circle, so except for the fact that 1,2,3 could have been along the upper curve of the dial it doesn't seem like an applicable metaphor.

Posted by Pierce on January 20, 2005 at 08:50 AM

"Rollovers" bugged me too (pun intended). I think it's a perfectly reasonable way to indicate that something can be clicked (the miniscule time difference is worth the added space and prettiness), and when used well almost certainly improves clicking accuracy.

"URL Naming Bug 2" is fairly ridiculous. I agree it's unintuitive to someone who has never used URLs but most computer users are either used to the way we do it now or will never have to deal with URLs past www.blah.com. Making any kind of change on that scale is crazy anyway. It's like switching to Metric. It's only more useful if nearly everyone makes that switch, both mentally and in infrastructure.

Posted by Vinnie on January 20, 2005 at 11:47 AM

Hm, I guess I will have to agree that rollovers are a good thing- seeing as how I have used them in the design of my new website. I think it was important to highlight the fact that the images on the side of the graphic were actually links and that the site was more than just a graphic.

I understand the fundamental complaint about the difference in keypads. But if you think about it, people don't type on the phone the same that they do on a keyboard. The phone (and also remote controls) are used by the user's thumb, whereas on the keyboard the user is using their fingers. Did you ever shop at Price Club when the cahsiers punched all the UPCs in instead of barcodes? They were wiz's on the keypad! I was definately impressed.

I have not taken the time to read his list, but I agree with all of your comments on this page. But have you tried the WindowsXP cd-write wizard? I found it works faster than Nero and didn't glitch. So the fact that it is implemented with the rest of the operating system as if it was another disk I think is actually working. True, people need to understand how they are different- but that will never fully happen with anything computer related. :)

Posted by Dewey on January 26, 2005 at 09:12 PM
Post Comment

Email Address: