The Top Eight Computer Myths!

This column originally published 1999,
in Australian Personal Computer magazine.
Last modified 03-Dec-2011.


Since most computer problems are resolved either by random or ritualistic action on the part of the user, or by following mysterious instructions from an allegedly knowledgeable support person, it's hardly surprising that there are lots of pseudo-religious myths about PCs. Between them, these stories can waste a lot of your time and money. Worse yet, you could become one of the myths' promulgators, and exacerbate the problem!

This article is partly about the worst of these myths, but it's also about how to spot misconceived ideas yourself, and where to find the straight dope about your computer.

1: Everything's a virus!

By far the most common PC myths concern viruses. Novice computer users are renowned for assuming that anything that goes wrong with their PC must be caused by a virus, and misguided virus warnings are one of the most popular kinds of non-commercial junk e-mail. Many of these warnings clearly originated as a joke (e-mails titled "Pen Pal Greetings!" that warn you never to open an e-mail titled "Pen Pal Greetings!", for instance...), but many computer users don't know what their computer can and can't do and so forward the "warnings" promiscuously.

In recent times, it has become possible for e-mail to be harmful. "Rich media" e-mail clients, especially those made by Microsoft, have far more complex abilities than client software like Eudora, that just tries its best to render whatever you've been sent, and doesn't automatically run anything. By means of legal function calls or unintended weaknesses, overpowered clients like Microsoft Outlook Express can do strange things when they receive particular messages, like for example an HTML e-mail message that's broken in just the right way.

Far commoner are e-mail pseudo-viruses like good old "Melissa", which is part Microsoft Word macro virus and part "Trojan". These pseudo-viruses pretend to be something harmless - an e-mail with an attached text file, a movie file, a list of smutty URLs, something like that, but aren't. Unlike a true virus, though, they have to be specifically executed before they do their thing. Melissa's more sophisticated offspring don't need Word to work - the various Loveletter variants, Serbian-Badman, Life Stages, for instance, are all self-contained.

Most of these viruses require the foolish user to manually execute an file, usually attached to an e-mail, but some of them (like Loveletter) can run if you just have the Preview Pane activated in Outlook Express. The preview helpfully executes the attachment; opening the e-mail normally doesn't. Fabulous piece of software design there, guys.

Microsoft released a patch on that dealt with this problem, by the way, but only after pretty much everybody running Outlook in their office had had their server jammed solid with virus attachments.

Originally, viruses that propagate via e-mail did it exclusively via Outlook. If you used a different e-mail client, you quarantined the infection to your computer.

More recently, worms and trojans have been coming with their own built-in mail sending software, and the ability to hoover up destination addresses (and fake "From" addresses) from various mail clients, Web browser cache files and other locations. If you're not running a Microsoft mail client, though, you still have to actually deliberately execute the virus program to kick things off.

On PCs, mail-client-exploiting viruses are likely to remain restricted to Microsoft software, partly because its excessively featureful nature and poor security make it a great target, and partly because it's so popular. Similar weaknesses have been demonstrated in Corel's SCRIPT files, and anything with a similarly powerful macro or script language is theoretically susceptible, but the victim populations for viruses that target other software are small enough, and Microsoft software still vulnerable enough, that there's no good reason for virus writers to bother with other strategies.

Avoiding viruses, in general, is fairly simple. Use virus checker software. Keep it updated. Don't run mysterious programs sent to you by e-mail. Look carefully at attached files to make sure that they're not called something like foo.txt.vbs - a Visual BASIC Script file, not a text file. Don't run macros in unknown documents in programs like Microsoft's Office suite. And don't use Microsoft e-mail clients unless there's some amazingly good reason to do so.

Viruses which are alleged to physically damage hardware are also hoaxes, although some come close. A virus like the still widely-found CIH, for example, can wipe many computers' BIOS chip. This is a Bad Thing if you've got a late model motherboard with a socketed BIOS, since you'll probably need to swap in a new chip, or at least get the old one re-burned by someone. It's a Very Bad Thing if your motherboard has the BIOS chip soldered onto it, since then the most economical solution is probably to replace the motherboard.

But this isn't, strictly speaking, hardware damage; the BIOS chip itself still works, it's just had rubbish data written to it.

Some really old computers (we're taking 1980, here) may or may not have had particular hardware oddities that might have allowed malicious software to damage hardware, by for example ordering the monitor to try to paint the whole image in one tiny spot. But anything modern enough to run DOS, let alone Windows, has no such problems.

If you've received a questionable virus warning, or if you just want to know how to spot them, the (formerly Computer Virus Myths) web site is the place to go. For information on what viruses really do exist, start at Virus Bulletin, the best respected independent virus information source. Symantec's AntiVirus Research Center database, searchable here, is another handy resource.

2: Microwave monitors?

Much has been made by some people - particularly those selling monitor radiation shields - of the supposed health risks of the emanations from computers in general and monitors in particular.

There is evidence that EMR (electromagnetic radiation) - specifically, the radiation from mobile phones - can have effects on living tissue beyond those expected from simple heating. But it most certainly does not follow from this that even high levels of mobile phone-type radiation actually cause any adverse effects at all in humans. Epidemiological evidence doesn't support such a conclusion, despite sensationalistic reporting of inconclusive studies. Since radiation from computers is both a great deal weaker and at very different frequencies from mobile phone radio waves, there is no reason at all to suppose that your PC is giving you cancer, or making you infertile, or whatever the panic-mongers are alleging this week.

Sitting in a bad chair typing frantically for hours isn't good for you, and if you spend all of your time in front of a PC instead of getting some exercise, your health most certainly is at risk. But since everybody already knows this and most people ignore it, it doesn't make the headlines.

3: Leaving it on

Some people leave their computers on as long as possible, on the grounds that components come under the most stress when turned off and on, and so you're less likely to have failures if you leave the power switch alone. There's only a grain of truth to this.

Rapidly cycling the power - working the power switch as if you're being paid by the click - is a bad idea for many electronic devices, including computers and monitors. If you're talking about ordinary use, however, the only problems you're likely to encounter stem from differential thermal expansion. Things get bigger as they warm up and smaller when they cool, and different components in a computer expand and contract by different amounts. The resultant mechanical stress can, theoretically at least, break traces on circuit boards and cause similar havoc.

In the real world, thermal problems with personal computers practically never have anything to do with differential expansion, but instead stem from lousy ventilation. Hot components, hard drives in particular, can barbecue themselves into an early grave. But these failures happen pretty seldom, these days, and modern hard drives are very unlikely to suffer motor or solenoid failure on startup. An old drive that's developed "stiction", where the drive has a hard time spinning up, should be left running all of the time. But that problem's never been common and is now close to unknown.

Current "green" PCs with power management features let you have most of the convenience of an always-on PC without the power bills; you can put your monitor, your hard drives, your processor and even the whole computer to sleep, and wake everything back up in a few seconds. From a differential expansion standpoint, this is the same as manually powering off the components in question. But since differential expansion is unlikely to ever do anything bad to your computer, who cares?

4: Screen savers

The screen saver is a modern art form. But what it isn't, any more, is a way of saving anybody's screen from anything.

In days of yore, monochrome monitors were quite susceptible to a phenomenon known as "burn-in" or "phosphor burn". An image shown on the monitor for a long time - a default menu, for instance - would burn in as an incurable ghostly image. Some colour monitors are still susceptible to this, but only if the image has been on the screen for a really, really long time, which is something that just doesn't happen in most applications.

All you need to protect your screen, of course, is to blank it; again, current PCs with their standby features make it easy to save electricity as well as the screen.

Some screen savers won't help with burn-in, anyway, because they have graphic elements that never move. If the screen-saver's static, it's as bad as a static application screen.

5: Don't defrag!

You don't need to defragment your hard drive very often. Modern drive optimisers like Windows 98's Defrag which position program data according to how often you use it can, indeed, improve performance a bit, but there's no reason for even a heavily used computer to be defragmented every week, or even every month. Yes, it'll be faster if you do. But the difference will probably be tiny.

Hard drive performance in toto makes very little difference to system performance, on machines with adequate physical RAM. The difference in performance between unfragmented and moderately fragmented drives is small, and the larger the drive, for a given level of filesystem activity, the less fragmentation it will suffer.

If you're using Windows NT or 2000 and NTFS-formatted drives, bear in mind that NTFS is famously insensitive to fragmentation - which is just as well, because it's hard to do anything with NTFS without it fragmenting data. This is why Microsoft claimed for so long that NTFS was immune to fragmentation, and no defrag utility was needed at all!

NTFS performs poorly on old drives with lousy seek speed, but the trade-off is that its performance as fragmentation increases remains quite steady. Once the NTFS Master File Table (MFT) becomes fragmented, you can indeed lose performance, but how much you lose still depends on what files are where and how you use the computer. Look at overall system performance, rather than just disk subsystem performance, and the difference due to fragmentation often fades into the noise.

How much effect fragmentation has on performance depends heavily on what files are fragmented, where the fragments lie, and what filesystem you're using. Generally speaking, the upshot of all this is that frequent ritualistic defragmentation, in the absence of a significant measured performance loss (not just how your computer "feels" to you), is, obviously, unnecessary.

Fragmentation certainly can severely degrade system performance, especially on Windows machines without enough physical RAM, or which are doing very disk-intensive tasks like serious database work or high data rate video editing. Defrag weekly, though, and you're probably just going to grow hair on the palms of your hands.

6: Monster swap files

Twiddling Windows' virtual memory settings (setting a static swap file size, for instance) doesn't improve Windows 98 or ME's performance much, if at all, although it can help Windows 95. If you choose to do it, don't use the goofy rule-of-thumb that your virtual memory should be some set multiple of your physical memory. You need as much memory as all of the programs you want to run at once will consume, and no more. The more physical memory a given system has, the less swap file size it needs, all other things being equal.

This goes for Windows NT and 2000, too; they let you specify swap file sizes on multiple different drives, which is nice to split swap activity over different physical devices for performance purposes. But going bananas on a multi-drive machine and giving yourself a permanent 3500 megabyte memory pool is pointless. Set your minimum total swap to roughly match your average memory pool needs, if you like (Ctrl-Alt-Del, Task Manager, Performance tab; the "Peak" number shows the most memory your system's used this session), but no tweaking beyond that is needed.

7: Microsoft bit rot

If you're running Windows 95 or 98, resign yourself to the fact that, over time, it goes rotten. Windows ME's "save point" automatic backup system may help, or it may not. Sooner or later, any Windows 95-series PC that's had programs installed and removed or its configuration otherwise changed a fair few times is going to start misbehaving badly enough that zapping the Windows directory and reinstalling, if not formatting the whole hard drive and starting afresh, is called for. Even WinNT and Win2000 systems will go bad eventually, given enough fiddling.

This does not mean you have a virus, or defective hardware, or bad karma. It's just one of those endearing character traits that make us all love Windows so much.

8: Magnetic mayhem

One of the basic tips in every computers-for-idiots book is to keep your floppy disks and other magnetic media away from magnetic fields, lest your data be wiped.

Magnetic fields can, indeed, eat data, but the usual suspects aren't generally the problem.

Every ordinary cone-type speaker driver has a large permanent magnet on its back, to give its electromagnetic voice coil something to push against. "Shielded" speakers neutralise the magnetic field with another, opposed magnet glued to the voice coil one, which reduces the driver's efficiency but does the job. An unshielded speaker's magnetic field is clearly apparent if you put it too close to your monitor - the image will distort and change colour, and in extreme cases stay that way, despite the monitor's built in "degaussing" circuit, until someone makes magic passes over the device with a degaussing wand.

But putting a floppy disk on top of an unshielded speaker won't hurt the data on the disk at all. This is partly because a change in magnetic field strength is helpful in erasing disks when the field is more or less powerful enough to flip the bits, but mainly because the magnet on the back of a speaker may only barely be strong enough to erase a floppy even if you put the disk right on top of the bare driver. A rapidly oscillating magnetic field of greater strength is much more dangerous than the weaker, static one from a simple permanent magnet inside a box.

A old fashioned bell-ringer telephone contains a pulsing electromagnet that moves its bell-clapper, and a disk leant up against it is likely to be a goner in short order. The abovementioned monitor degauss circuit may zap disks left atop the screen pretty well, too, and some printers, especially older dot matrix models, contain quite large and poorly magnetically shielded motors.

Essentially, any electronic device with moving parts may be able to zot your floppies, tapes and Zip disks. It doesn't have to be able to pick up a paper clip to be dangerous.

Data sources

For the good oil on your PC on the Web, check out the PC Webopedia, the Free On-Line Dictionary Of Computing, and the magnificently complete PCGuide. The first two, between them, can define practically any computing term there is, and the last one is a very well done, reliable and up-to-date reference on PC hardware and software in theory and practice. If you want a paper reference, a monster PC databook like Scott Mueller's excellent Upgrading and Repairing PCs is another great way to immunise yourself against computer misconceptions.

Give Dan some money!
(and no-one gets hurt)