Microsoft Windows, updates, reboots, and the principle of least surprise

It’s been a while since I’ve ripped on Microsoft, which I was reminded of after seeing Matt Mullenweg’s recent blog entry about Windows and it’s horrid habit of rebooting to install updates, without specifically being told to do so. This wouldn’t be nearly as bad, if Matt didn’t link to a blog entry he wrote six years ago describing the same exact thing which at the time cost Matt quite a bit of unsaved work. (At the time of this earlier entry, Robert Scoble was working for Microsoft and apologized to Matt in his blog. By the way, this is something I suspect would never be allowed from a Microsoft employee now, and looking back at it, I’m a bit surprised Robert got away with it.)

Now, most people would think Microsoft would listen to their customers(*) and fix what is widely perceived as a bug, even if it was originally an intentional design decision. Especially after someone as high-profile as Matt has blogged about it, now twice. For the uninitiated, here’s what Matt is talking about in those entries:

  • The Windows Update tool downloads its updates and installs them. There are three options: manually download and install, automatically download and manually install, or automatically download and install.
  • At the conclusion of the install, if one of the updates requires a reboot (which, given this is Windows we’re talking about, is almost always the case), a dialog box pops up with the option to “restart now” or “restart later” with a timer. Originally, this timer was 5 minutes; I think it’s been changed to be longer (15 minutes?).
  • “Restart later” just means the dialog box pops up again some time later. I’ve never measured how long it is, I just know it keeps nagging until you reboot, and there is no way to select “go the #%&$@ away, I’ll reboot when I damn well please.”
  • “Restart now” does what it says.
  • If no action is taken when the timer expires the system reboots. This is almost never what the user (who is in effect Microsoft’s customer) actually wants.

This last bit is why people like Matt write blog entries like that one. This is also why people like me quit running Windows at all. My first choice for a new PC is “one without Windows on it.” My second choice is “one I can wipe the Windows install from as soon as I get it.” (I’ve been stuck using a Windows XP system enough to know this is how it worked as of that version. I quit using Windows at the first opportunity, for reasons that should be obvious. No PC that I actually own has run any version of Windows since 2002 April, which is now over eight years ago.)

The reason the timed reboot is such a disaster is that it violates the principle of least surprise (also called the principle of least astonishment). A computer user expects the computer to stay running absent a specific command to reboot. (I’m being generous here, given this is Microsoft Windows in question, an operating system not exactly known for its stability.)

In addition, Microsoft puts scary warnings in the Windows Update component which encourage users to select “automatically update.” From Matt’s post:

I can’t reconcile that it was due to a feature of an operating system, a feature I was told to turn on to stay safe, and a feature that bugs you when it isn’t activated. I trusted the computer because of the improvements to stability Microsoft had made in XP and SP2. Trust like that is slow to build and easy to break.

Now, Ubuntu gets it right (or at least did as of 9.10; I lack the experience with 10.04 to confirm). Most updates do not require a reboot. Those that do (new kernel versions), pop up the dialog box once. Yes, that’s right, once. If you say you would prefer to reboot later, the dialog box is gone for the rest of the session. The only bad part of this is that hibernating is broken from that point forward until you reboot and you just have to know this from experience. Since Ubuntu’s Update Manager is relatively unobtrusive (it pops up once per day if dismissed) this effectively becomes a non-issue. If you don’t use hibernating mode it’s not an issue.

Anyway, my point is that it’s inexcusable for Microsoft to allow what most users consider a horrid bug in place all this time, especially given a competing operating system shows how it should be done. I don’t know the right way to fix this in Microsoft’s world. I do know defaulting to a reboot after a time delay is not an acceptable answer to the majority of computer users (note, that’s computer users in general, not just PC users or Windows users). Then again, maybe the answer lies in not requiring a reboot after just about any system update. I’m not holding my breath waiting for Microsoft’s fix for that one, though.

(Note: When I refer to Microsoft’s customers here, I refer to the end users of Windows. I know technically Windows is usually sold to OEMs like HP, Dell, Compaq, Gateway, etc. but they aren’t really the customer here. It is the end user that Microsoft ultimately aims to please.)

Windows 7 boot time shenanigans

According to a recent CNet article, it seems that Microsoft has been a bit deceptive with their claims that Windows 7 boots faster.

The claim is from a company called Iolo Technologies:

[Iolo’s] lab unit found that a brand-new machine running Windows 7 takes a minute and 34 seconds to become usable, as compared to a minute and 6 seconds for Windows Vista. Iolo notes that it measured not the time it takes for the desktop to appear–which can be as little as 40 seconds on a fresh installation of Windows 7–but rather the time it takes to become fully usable “with CPU cycles no longer significantly high and a true idle state achieved.”

I’m not the least bit surprised that Microsoft would take the deceptive, underhanded path here, and make Windows 7 look like it boots faster even while the rest of the “booting” is still going on in the background making the system relatively unresponsive. Much of the rest of the computer-using public, however, falls for this kind of thing hook, line, and sinker.

I have no access to a computer running anything more recent than Windows XP Media Center Edition, so I cannot unfortunately lend my personal insight there. (Some of that is by choice: I hopped off the Windows train at Windows 98, and my next new PC will come preloaded with a GNU/Linux distribution called Ubuntu which is an offshoot of Debian. I’m running Ubuntu on this rather geriatric PC (800 MHz Celeron, 256M RAM, 20G drive) that I am using to write this, and if I’m careful about what I run it’s not too bad. Certainly much better than the Windows 98 it left the factory with.)

What I can tell you, is that with just about any operating system and GUI released as free software, what you see is what you get. The desktop or login screen comes up, and that means the system is done booting. Microsoft would do well to adopt the same model of transparency, or drop their deceptive claim that Windows 7 boots faster when in fact it probably does not.

I am definitely curious as to how fast Ubuntu 9.10 (Karmic Koala) and Debian
5.0 (lenny) would boot on the same hardware Iolo Technologies used for their
test. I have never actually timed the boot procedure on this PC, but I have
nothing to really compare it to so it may not be that relevant.

Microsoft and News Corp: strange bedfellows at their worst reports on a very troubling development. Microsoft and News Corporation are forming an alliance against Google (if you believe them) or against all of us (if you don't). In essence, News Corporation will "de-index" its content from Google in exchange for payment from Microsoft.

At best, this deal is dubious. At worst, it's anti-competitive and forces the users of search engines into a chess game as unwilling pawns. My argument here is not in favor of Google any more than it is in favor of choice of the users and against underhanded tactics on the part of Microsoft.

I can't see the US Department of Justice just letting this go unchecked. I'll just say there's never been a better time to boycott Microsoft than now.

To be fair about it, we do need more choices than Google or Bing. Cuil showed some promise at first (once their bot became better behaved). Several others have come and gone before the age of Google (I still miss many of the searches AltaVista let you do that Google simply does not have functionality for).

In a way, I had hoped Bing would become a serious alternative to Google, just to keep Google honest. I should have known better than to think honesty and anything related to Microsoft belonged in the same thought pattern. Shame on you, Microsoft.

Astroturf a la Redmond: Windows 7 Parties

While cleaning out the draft posts queue, I found this. The original article is a bit old, but the parties haven’t happened yet.

A recent TechFlash article discusses a Microsoft initiative for the upcoming Windows 7 release, describing it as a “Tupperware-style twist.” The idea is to encourage users, partners, and of course Microsoft employees to throw parties to show off Windows 7.

I see, as the title implies, what is essentially astroturfing at its worst. If Windows 7 were that great of an operating system, Microsoft would have people volunteering, or even paying Microsoft, to have these launch parties.

As far as my personal PCs go, I haven’t really looked back since the spring of 2002 (I didn’t write down the exact date, unfortunately) when I reformatted two different Windows 98 PCs and installed, at the time, the GNU/Linux distribution maintained by Red Hat on one (today’s equivalent would be the community-supported Fedora Project), and FreeBSD on the other. (Both of those PCs eventually wound up running Debian GNU/Linux years later.)

I do use a Windows XP laptop PC, which still has what I deem to be an unacceptable crash rate. By unacceptable, I mean it crashes at least five to ten times as often as the Debian GNU/Linux PC next to it which is at least most of a decade old. No, that’s not a mistake. Simply put, even with less than half the CPU and a fourth of the RAM, I get much better stability, even with the obvious reduction in performance. Put simply, Centerpoint Energy (our local electric utility) and the forces of nature responsible for thunderstorms force reboots of the Debian PC more often than any technical problems with Debian itself.

(Why haven’t I bought Apple’s products instead? Regular readers should know this, but I’ll provide a starting point for the new readers.)

Microsoft has still done next to zilch with regard to helping ensure the freedom of its customers. In fact, Microsoft has pretty much made itself the sworn enemy of the free software movement, with apparently no shame or regret. While Microsoft has made token efforts to contribute to the open source movement, it is very important to note that the ideals of the open source movement only encourage access to the source code for convenience and open source licenses do not always protect all the essential freedoms of users and programmers of the software released under them. It is also important to note that without the work of Richard Stallman and the FSF on the free software movement, there would be no free software movement for the open source movement to have splintered from.

Those readers unfamiliar with what I discuss above are encouraged to read some of Stallman’s essays, most notably these two: “Why Open Source misses the point of Free Software” and an earlier version “Why ‘Free Software’ is better than ‘Open Source'”.

Several of Microsoft’s licenses appear to have been intentionally worded to provide the illusion of freedom while in reality providing just the opposite. Most notably, these are the Limited Public License (Ms-LPL) and Limited Reciprocal License (Ms-LRL). Both of these licenses require that any modified versions of the original code must run on Microsoft’s Windows operating system. To those who value freedom on their own terms, not those of a large corporation with no particular incentive to be nice, this type of restriction is abhorrent.

In summary, my view is rather simple. Windows 7: same song, seventh verse, even bigger and even worse.

“Library of future” initiative becomes corporate battleground

Wired reports on Sony’s decision to side with Google in a highly contentious lawsuit between Google and rivals Microsoft, Yahoo, and Amazon.

The lawsuit centers around privacy concerns and the fact it would give Google monopoly-like status on book rights that would be impossible for other companies to acquire without their own lawsuit.

Worse for Google, the Department of Justice is also investigating the settlement–a rather ominous and foreboding development.

I have never been all that positiviely impressed with Sony; they are probably the only company with a hand in consumer electronics and entertainment (the latter through their acquisition of Columbia Tri-Star in 1989 and CBS Records in 1987). The second DVD player my mom ever bought was a Sony, and it was the first to fail; the RCA player purchased a few months before still works today as far as I know. It has always seemed to me that Sony built up a good reputation in its early days, and somehow managed to keep it afloat enough to justify some kind of premium pricing even though the reputation it has is probably less deserved today.

Still, today, I’d really like to give Sony the benefit of the doubt. Yes, even though this is the same Sony known for the doomed Betamax and Digital 8 videotape formats, and the XCP and MediaMax copy protection scandal of 2005.

I don’t know much of the details and motivation behind why Sony would back Google. I do know that it’s Very Bad to let any one company grow to an effective monopoly; there is a reason we have the Sherman Anti-Trust act in the US and why similar legislation and oversight exists in the EU and elsewhere. And this does smell like something Sony would do not out of concern for its customers, but for its own corporate interests. I also believe we, as a society, should not reward a company that puts shareholders above customers when filing amicus briefs in these legal chess games.

Maybe my instinct is off the mark yet again, but it is what it is.