Welcome, Digg visitors. Wow, twice in three days an old post of mine gets picked up and Dugg like crazy. Just to be clear: If you have a specific problem with removing a specific program, a registry cleaning utility might be able to identify keys that will help you solve that specific problem. But that’s a rare scenario. Most people I know use registry cleaners as part of their magic cleanup routine, and I see very little upside and a lot of potential downside in this sort of routine use. Specifically, as I write below, I have never seen any evidence that routine “cleaning” of the registry has any positive effect. I stand behind that statement.
Via Matt Goyer, John Hoole offers this cautionary tale:
just a note to say if you have Windows XP Media Center Edition 2005 (probably all versions actually) steer clear of registry clean programs such as Reg Mechanic they go through your registry and delete unnecessary keys….. sounds good but it didn’t count on Media Center I ran it a few days back and when I came to use Media Center it loaded then produced a crash report and died, took me ages to figure it out until I came to run Reg Mechanic again and realized This program deletes DLL files too so….. I restored the first backup and rebooted and media center worked fine so if you have that error on startup that’s your problem right there. Just restore the backup from Reg Mechanic. So you have been warned.
I’d go a step further: Don’t run registry cleaner programs, period. I won’t go so far as to call them snake oil, but what possible performance benefits can you get from “cleaning up” unneeded registry entries and eliminating a few stray DLL files? Even in the best-case scenario the impact should be trivial at best. Maybe a second or two here and there, maybe a few kilobytes of freed-up RAM, and I’m being generous. How can you balance those against the risk that the utility will “clean” (in other words, delete) something you really need, causing a program or feature to fail?
If anyone has done any serious performance testing on this class of software, I’d be interested in seeing it. In the absence of really rigorous testing and fail-safe design, I say: Stay far away from this sort of utility.
If you have a counter-argument to make, leave a comment. But simply saying, “I use Reg-o-matic Deluxe and my computer is way faster than ever!” isn’t good enough. Show me the data!
Update: I did a Google search for “registry cleaner” performance tests, and got more than 25,000 hits. In the first 15 pages, however, there wasn’t a single example of an actual performance test. Virtually all the results were from companies that make and sell this sort of utility, or from download sites that have affiliate agreements with these developers. I found one recent how-to article from Ed Tittel on TechWeb. Ed asserts that “Most Windows experts recommend a Registry clean-up on all systems at least once every six months.” He didn’t link to any of those experts, however.
Later in the same article, Ed advises: “I urge you to check comparative reviews, ratings, and rankings of Registry Clean-up Tools before you invest hard-earned dollars on these products.” Sadly, there are no links here either. I suspect that’s because detailed comparative reviews of this class of software don’t exist. Ironically, the article inadvertently documents the case against this sort of utility. Early on, it states: “The typical Windows system has literally hundreds of thousands of Registry entries.” The screen shot from the free utility he spotlights shows a grand total of 19 “errors,” most of which are simply pointers to CLSIDs that don’t exist. Is it really worth spending hours on this task? I don’t think so.
The best bit of reading I found in my search was this rant from a poster named Jabarnut on a thread at DSL Reports’ Software Forum:
The Registry is an enormous database and all this “Cleaning” really doesn’t amount to much…I’ve said this before, but I liken it to “sweeping out one parking space in a parking lot the size of Montana” … a registry “tweak” here and there is desirable or even necessary sometimes, but random “cleaning”, especially for the novice, is inviting disaster.
I also would like someone to show me any hard evidence that registry cleaning actually improves performance. (Unless there is a specific problem that has to be addressed by making changes to the registry).
Sorry to go on like this, but I feel there is way to much Registry “Cleaning” going on these days just for the sake of “cleaning”.
Amen.
Update 11-Sep: Several commenters have made a good case for a handful of utilities that include registry repair and cleaning options. They make the point that these are useful when used intelligently, not indiscriminately. My colleague George Ou from ZDNet passed along these comments:
I do like the free CCleaner. I’ve cleaned out 1 GB or more of junk on friends computers and it does make the system a little more responsive. You don’t get as many unexplained pauses. This is a problem with the lack of multithreading in Windows Explorer most of the time when it times out on dead resources like a detached network drive. I thought I remember reading something on the Vista features that fixes this by supporting multiple threads.
Other than that, I’ve made sure that I don’t have any dead links the system is trying to access on the desktop that are sure to cause a 30 second lockup even if I drag an icon across the dead link icon. Ccleaner also does a nice job removing a lot of that junk. The combination of MSCONFIG and Ccleaner works wonders.
OK, I’ll give it a try.