David Pogue wrote an article that was published in the New York Times today entitled "Want a New Headache? Try to Uninstall." (no link, because the NYT discourages it) in which he complains about how difficult it is to uninstall programs. It may not be obvious why until the end of this post, but I think David Pogue is not qualified for his position as a "technology columnist". He writes:

The first program I spotted was Adobe Acrobat 5, which I don't need any more because I now have Acrobat 6. But when I tried to remove Acrobat 5 (using Windows?s Add/Remove Programs program), a message said, "The system indicates that the following shared file is no longer used by any programs and may be deleted: C:/program Files/Dell/ShareDLL/djbsdk.dll. If any programs are still using this file and it is removed those programs may not function. Do you want to remove the shared file? Yes/No." WHAT THE?!?!? Like I'm supposed to know if some other program is going to need C:/program Files/Dell/ShareDLL/djbsdk.dll?

No normal person could possibly know that—heck, no technology columnist could possibly know that—which can mean only one thing: That whoever writes that kind of error message is lazy and mean-spirited.

While this message is indeed annoying, Pogue's reaction to it is unreasonable. When he writes "No normal person could possibly know that", the "that" probably refers to "the answer to the question being asked", and the question being asked is "Do you want to remove the shared file?" I think a very normal person, perhaps slightly more intelligent than Pogue, could figure out "Well, it actually says in the message that if I remove it, programs might cease to function, so it's probably more conservative to say 'No, I don't want to remove it'". Upon thinking about it a little bit further, the person might realize that "djbsdk.dll is in a directory owned by Dell, who manufactures my computer. They have nothing to do with Adobe Acrobat. I am still using my Dell computer, so in fact, I'm certain that I should not delete this file."

If that person were able to think from the error message writer's perspective, he might realize "As an error writer, I know that the software is in the position of wanting to delete a file, but realizes that for some reason or another, deleting this file might be a mistake. The software wants to check with the user if it's okay to delete the file. How can I express this in such a way so that the user understands all the repercussions of his or her actions, and yet keep the message terse enough so that users don't simply ignore it?" I'd like to hear what Pogue's suggestion is for a better message (obviously, he just complains about the solution other people have come up with, rather than providing a constructive alternative).

Let's move on to program #2, from Creative Labs. This time, when I clicked REMOVE, it whirred for a moment and then said: "Do you want to completely remove the selected application?" Ummm, no. I actually clicked REMOVE because my index finger had an itch. DUH!

A lot of research has been done on how often confirmations should be asked. The general consensus is that if an action is irreversible (or very difficult to reverse), a confirmation should be made. So, for example, Microsoft Windows shouldn't ask for confirmation when you send something to the recycling bin, but it should ask for confirmation when you empty the recycling bin (in actuality, Windows XP at least asks for confirmation in both cases). Removing a program is generally considered irreversible, and so the confirmation is warranted. Furthermore, I don't know how much experience Mr. Pogue has in using the "Add or Remove Programs" utility, but did he notice that this utility is also used for changing the installation type of the program? If you've installed Microsoft Office XP, for example, and then click on the "Change" button, you can choose to add support for Japanese and other Asian text, if it wasn't already installed, for example. Notice how some of the entries in the program list have "(remove only)" after their names? The extra confirmation, "Do you want to completely remove this program?" is for people who click buttons without reading the label, because they assumed that every program in the program list would have the same button layout, and accidentally click on "Remove" when they meant to click on "Change". Given that there's 3 possible button layouts ("Change" and "Remove", "Change/Remove" or just "Remove"), I think this confirmation window is very well warranted. Also, simply clicking by mistake because your finger slipped is not unheard of. I've done it a couple of times before. That's three arguments, one backed by research into user interfaces, against Pogue's position. I think his argument merely consists of sarcasm, and um... well, implying that it's stupid.

Pogue goes on to complain about Firefox 0.8, Virgin Player, and a few other miscellaneous programs, and concludes with the following paragraph, which I've broken up to interject with my comments.

If all the best and the brightest programmers work at Microsoft, you'd think they could answer questions like these: If I click Remove, why must I be asked twice more if I want to remove something?

Regardless of whether or not all the best and brightest programmers work at Microsoft, they are able to answer that question. See, for example, SAP Design Guild for people-centric design's comments on confirmation dialogs: "Error prevention comes first [...] Prevent disastrous actions [...] If users might lose data display confirmation dialogs." Next question, Mr. Pogue?

Why can't Windows keep track of which programs need which pieces, so Microsoft's long-suffering customers don't have to judge whether some shared DLL file is still necessary?

There are two possible ways to do that. The first solution is to have the programs state what pieces they need. Microsoft could create a new standard which forces developers to tell Microsoft Windows what pieces they need whenever they install themselves. The developers would immediately cry bloody murder and refuse to use this feature, as all it does is creates unnecessary extra work. To "encourage" developers to use this feature, Microsoft released Windows DF, the next version of Windows, which will not allow a program to install itself unless it explicitly reports what files it needs to work. The end-user realizes that all his old programs, written before Microsoft invented this standard, do not work with Windows DF. The end users cry bloody murder. Hmm, this solution doesn't look feasible.

What's the second solution? Instead of having the software developers do the work, have Microsoft do the work: Microsoft could require all companies that produce programs to hire a certified Microsoft Program Removal Utility Quality Assurance Agent. The MSPRUQAA's job is to snoop in on all software written by all companies in the world, and then send reports back to Microsoft on what files these programs need. A company is not allowed to ship a product until the MSPRUQAA has personally gone through every line of code, making sure that a needed file wasn't overlooked. Furthermore, the company in question has to pay the MSPRUQAA a salary for the privilege of making sure their programs are compliant. This also means that hobbyist programmers like me (who cannot afford to pay the salary of an MSPRUQAA) will have to give up writing software for Windows. Hmm, this solution doesn't sound too hot either, and will probably result in some sort of anti-trust-Big-Brother-Microsoft-is-watching-everything-I-do lawsuit. What's your next question, Mr. Pogue?

When we install a new program, why aren't we asked if we'd like it to replace the older version, rather than making us mop up afterward?

See the above. There either has to be a standard way for a program to announce that it is a "better" version of an "older" program, or Microsoft has to have an agent in every company so that they themselves can judge which programs are better versions of which other program. This also adds the whole big mess of "Internet Explorer is a 'better' version of Firefox, so when you install IE, it automatically replaces the older Firefox for your benefit."

Of course, you already know the answer. Microsoft doesn't improve this kind of thing because it doesn't have to. It's got a bad case of a little thing called Monopoly Complacence.

I'm not sure if I should think that it's funny or scary that Pogue is this stupid and yet has a job writing the technology columns for the New York Times. Rather than making an constructive, or even informed complaint about the state of technology, he figures "Bashing Microsoft is hip these days, right?" and finds some random problem with his computer, and concludes with "Well, that's what you get when Microsoft is a monopoly." That's right, Mr. Pogue. It's Microsoft's fault that the beta version of Firefox had a bug in its uninstall program. What an idiot.