I have, for many years, always been suspect of reviewers of software. Most of  these well meaning reviews are based on a very short time for impression, and very few chances to see what possible interferences have been caused by the installation.

In the PC world, this is where the biggest problems occur. Too many software authors start with the premise that no one will be using anything but software from them, and extrapolate from there. [People with math backgrounds are keenly aware that, by nature, extrapolation leads to error.] This is becoming less prevalent, but still occurs frequently. 

If a program works well, but causes other, more important software to crash, or otherwise become unusable, where is the net gain?

In the Windows world, the next thing to worry about is what could be viewed as insidious, but frequently occurs without any thought by the author.  This is when an application installs, and replaces a Dynamic Link Library, with a newer, or sometimes older, copy by the same name.  When a function relied upon by another program no longer works, it can be hard to track what the problem is.  Most programs, but not all, ask when doing this replacement.  What happens then can be anybody’s guess, as the expected behavior is frequently not what happens. One scenario goes like this:

You install a program, it works great, you close it, and use another program, which happens to use the same DLL that the newly installed program uses. That older, already installed program, which happened to be minimized, did not allow the DLL in memory to be overwritten. The older program works just as it should…this time.  Tomorrow, after you have rebooted your machine, or shut the machine down, and then restarted, has now seen that DLL replaced, as the offending program had placed the different version DLL in a queue, to be replaced, as soon as the first was dropped from memory. Now the program you’ve had working for months suddenly goes haywire, and you can’t figure out why, as you haven’t changed it. Your machine has not had a catastrophic failure, and your disk drive is still ok. What gives?

Few reviewers keep from sending their analysis to the editor long enough to note and report this.

The next problem might actually be the worst. Since the same piece of software might get reviewed by many, over the course of its life, and changes in newer revisions might slip by. The reason they might slip by is due to human nature. When someone gets paid for their reviews, day after day, the chances that something might slip by gets higher.

A case in point:

The software firewall, available for free [which in itself tends to make reviewers more forgiving] Comodo, is a decent firewall, and gets rave reviews from those who think that not leaking is the be-all-and end-all. It is not. The company’s support pages are filled with strange incidents describing, in great detail, how this ‘learning’ firewall refuses to learn, and keeps asking, time after time, if an application should be allowed to access the ‘net. Further are the pages describing how, people with above average abilities, are completely stymied by the attempt to remove this program. Page upon page of descriptions of computers trashed by the removal of the offending bits left behind to mystify and vex the handiest of users. Yet the accolades continue.

Perhaps the best reason to be suspicious is a personal one. Their computer is not my computer. This is why the term Your Mileage May Vary was invented. As anyone from the EPA will tell you, the mileage reported is just pie-in the-sky.

Tags: dll hell, software reviews, comodo, ymmv