Dan's Data letters #39
Publication date: 27-Mar-2003.Last modified 03-Dec-2011.
Fudging the figures
Have you ever written an article about the appalling misuse of graphs in many benchmarks, often by people who should know better?
I couldn't spot one in your article archive and think its something more people should be made aware of as companies and reviewers use such tactics to mislead their audience by leaving axes unlabeled and not starting bar graphs from a zero origin. Some years ago I started taking Tom's Hardware a lot more seriously than most other review sites because their graphs ware always done properly and therefore didn't mislead (although their conclusions derived from these graphs sometimes do).
What prompts me to ask you this is this page at the VIA support site, which frankly makes me question my judgment in using a VIA chipset MB when they will resort to such grossly misleading graphing practices and statements. Can I trust a company that proudly proclaims a huge performance gain of about 250% according to the graph bar sizes, where the actual gain is even less than the statistical noise level, I would think?
I reckon it's about time some of these sites were taken to task about this and people were made more aware of how to spot a "marchitecture graph".
Lawrie
Answer:
I've railed against such things a few times, though I haven't done a separate
piece about nonsense graphs.
My most recent whinge was in this review, but this one has something to say about it too, and there's a passing mention here. This piece is also somewhat relevant.
Hoping for colour
Do you know if its possible to transform a black and white laser printer into a colour printer? Is it as simple as chucking in a colour cartridge? I'm guessing not...
Paul
Answer:
Colour lasers use a multi-toner system, which is a lot more complex, which
is why they're so much more expensive than the regular kind of laser printer.
You can get coloured toner for some black and white lasers, but it's just one colour, instead of black. Buy a red cartridge, for instance, and the computer and printer still think they're printing black, but they're actually printing red.
You could, in theory, do multi-colour documents by doing a colour separation on the computer, then feeding the printout through over and over and changing the toner colour appropriately (or by using multiple printers each with a different toner colour). But since repeat-feeding laser printouts is a recipe for crap all over the fuser roller, and also since the toner won't mix colours but will just print the latest one over the top of the previous ones, I doubt this'd actually work.
(You could partially solve the toner-on-top-of-other-toner problem by making it a halftone colour print, with different screen angles. This is how ordinary newspaper and magazine colour printing is done. The inks they use don't cover each other, though.)
Old with new
Can I use a GeForce 256 DDR graphics card in a motherboard with an AGP 3.0 slot? There is nowhere I can find that gives me a simple yea or nay. I'm upgrading, and I'd rather not have to fork out the cash for a graphics card at the same time as new motherboard, CPU, memory, PSU etc.
Markus
Answer:
It should work fine.
AGP uses a physical card-to-slot keying system which is meant to make sure that cards that need one AGP signalling voltage can't be plugged into a slot that can only deliver another one. All GeForce 256 cards should be 1.5V-capable, though (as opposed to the old pre-AGP 4X 3.3 volt signalling), which means you shouldn't have any trouble using them on current boards.
There are a few mis-keyed video cards out there which are 1.5V-only but can be plugged into a 3.3V-only motherboard, with disastrous results, but you don't have a 3.3V-only motherboard, so there's no way that can be a problem for you. I don't know whether any GeForce 256 boards were mis-keyed in that way, anyway.
Note that the signalling voltage is not the same as the main card chipset supply voltage; that's still 3.3 volts, no matter what kind of AGP card you're using.
Search for illumination
I am asking for an opinion about two LED torches - the CMG Reactor 2AA and the Streamlight 68202 4AA torch with seven LEDs.
Which one of these flashlights would be best suited for general emergency work? Work like changing flat tires, emergency car repairs, and situations of this kind.
The CMG Reactor seems to have a long range focus effect. The Streamlight 68202 has a rating of Class I, Division I hazardous type features for repairing a motor vehicle with a broken fuel system, for example.
I use a Brightstar Model 2206 six volt lantern for longer range uses. It is a class I, Division I hazardous type lantern.
Which LED torch is the most durable and sensible unit?
Tom
Answer:
The 2AA Reactor is a good flashlight for the money; I review it
here. I haven't seen a Streamlight 68202 yet,
though, so I have no firm opinion about it. It's bigger than the Reactor,
and would definitely be brighter than the Reactor running from alkaline
AAs - but the Reactor's meant to run from lithium AAs, and it's plentifully
bright then.
The Reactor is not a focusable or narrow-beam light, by the way; it's got the same not-very-narrow beam as various other Luxeon Star/O flashlights. It's got a beam width similar to that of a standard 51mm halogen downlight.
For emergency repair work, I personally would recommend a headlamp. Getting a flashlight aimed properly, and shifting it around when you need both hands for the work, is a pain.
The FrontaLED headlamps (as reviewed here) are excellent, but expensive. The Lightwave Illuminator's simple but serviceable; the Petzl Tikka is tiny, but not terribly bright; I check them both out here.
The Photon Fusion, as reviewed here, is a good multi-role light with decent brightness, but it's another fairly expensive option.
Photo power
I would like to ask your advice on batteries for use in digital cameras. I'm currently only getting about two hours of use out of my Canon A200 using the somewhat pricey Energizer E2 alkalines. I've noticed that you appear to know, well, quite a lot about batteries but I haven't really noticed any recommendations regarding what's a solid set of rechargeables for digital photography use. I'd very much appreciate your insight.
Dan [no relation]
Answer:
No-brand nickel metal hydride AAs are the best value. You can get brand
name NiMH AAs with capacities up around two amp-hours (2000mAh), but you'll
get much better capacity per dollar if you buy brandless 1500 to 1800mAh
cells. Good electronics stores have them; here in Australia, try
Jaycar. A dozen 1650mAh AAs from
them will cost you less than $50.
I'm not terribly enthusiastic about Jaycar's NiMH chargers, or those I've seen from other local electronics stores; they've got lots of fancy flashing-lights chargers with automatic or manual dischargers built in, but discharging isn't actually something that there's much point doing to NiMH cells, or NiCd ones for that matter. There can be some reason for it in very high current applications (electric model planes, and such) and now and then to equalise cells. But it's really not very important for normal users, and you certainly don't want a charger that discharges every time. That'll just wear the cells out early.
Of the Jaycar gear, I think their MB3535 charger's the best of a bad lot; you'll need to spend more money on a plugpack to use it from mains power, though.
I'm reviewing a Maha MH-C401FS charger at the moment; it's probably the best consumer 4-AA charger in the world right now, and can be had with an Australian plugpack (it's another 12 volt unit, so getting an American C401FS working from Australian mains power would just require an Aussie plugpack). Very fast, an individual charge circuit for each cell, and charges AAAs as well.
Unfortunately, there's no Maha distributor in Australia at the moment, so you'd have to buy from overseas, or try eBay.
[Now, years later, there is an Australian Maha distributor - Servaas Products. They have decent prices, and of course also cheap delivery for Australian customers.]
A reader's kindly informed me, now, that you can get Rayovac's current super-fast charger models in Australia now, and that they work well. Dick Smith Electronics has them (catalogue number S7811) for $AU79.96.
Watching pr0n for a living
In relation to your article on censorware, I believe there is a company called N2H2 offering a filter that is based entirely on a human reviewed list.
I am not sure if it works more or less effectively that the others you have reviewed, but it does appear to contradict your claim about whether the human review thing can be done. It also raises some very amusing workplace health and safety issues for the presumably large number of people who must be employed to ceaselessly scan the web for smutty material to add to the blocked list.
Stefan
Answer:
I actually mention N2H2 in my censorware article.
As you say, they claim their list is "human reviewed", but I don't think
they say anywhere that every URL on it has been "human reviewed".
I think they human-review sites that're flagged by their robots as being
in generally objectionable categories. There's so many URLs out there, and
so much change day to day, that I think it's going to remain impossible
to human-review every URL in a properly comprehensive database, without
employing thousands of people.
If you only bother with domains, and categorise every sub-page identically, then it's doable. There are more than 30 million domains registered at the moment, but if you've got a hundred people categorising them and they each do, say, one domain every 30 seconds, six hours a day, then you can get through the lot of them in a year and a bit. N2H2 have been around long enough that I can believe they could have pretty solid human review of every domain out there. I wouldn't be a bit surprised if they didn't, mind you; it's still quite possible to find naughty stuff, on sites that've been around for a while, which they don't know about. And they still block several anonymisers and translators, under the "loop hole" category. Single domains with lots of different content on them (geocities.com, for instance) considerably inflate the number of pages they'd have to deal with in order to come up with a comprehensive database; apparently they just put all of those sites in the "free sites" category, which many school users of N2H2's Bess system block.
Since N2H2 seem to have pretty much squashed the bugs that a Web search for complaints about them over the last few years will find, they do appear to more or less have it together now.
Who knows, though? If you decrypt their database to see what you're buying, they'll probably sue you under the DMCA.
From small to slightly bigger
I have a problem with a new hard drive I just installed. I had a 2.4Gb drive and I bought a 4.1Gb; I've installed the 4.1 (all up and running A-OK) and formatted the 2.4. When I got the new hard drive there was a file called COMMAND already on it; I assume this is a factory installed file so I copied it, believing I would probably need it when format of the 2.4 was complete. The only problem is I haven't got the foggiest how to re-install the COMMAND file back to the 2.4. Also, once this file is re-installed, is there anything else regarding this change around that I should be aware of?
Nick
Answer:
COMMAND.COM is the basic MS-DOS system file, but it's not the only one.
More are needed if you want to boot DOS from the disk. None are needed if
you don't.
If you're running Win98SE or earlier, or have some other way to boot DOS (like a boot floppy), you can write the MS-DOS system files to any disk you like by going to the DOS command prompt and typing
SYS [driveletter]
So, for instance,
SYS D:
will write the system files to D: (assuming D: is your second hard drive and not your CD-ROM drive, that is).
The scratch mix
I can't get any definitive advice on polishing out a couple of scratches my monitor recently suffered on the way back from a LAN. I have been recommended all sorts of things such as toothpaste, but that appears a little too rough for my liking. After lapping many heatsinks, I realise that a better finish will be achieved with more patience and a finer grit, but I'm not willing, nor do I consider it a good idea, to polish my monitor with sandpaper of any sort.
I have been pointed in the direction of "Meguiar's Mirror Glaze 17 Clear Plastic Cleaner", but am unsure if its just going to be a waste of $AU20 for a bottle. Another recommendation has been for a product called "EEE" or "Triple E", but apparently that works best on metal.
The monitor in question is a Philips 109S and I've only had it for about 6 months. I don't think it has any protective coating on it, but am unsure.
Does Philips or any monitor manufacturer have a repair centre for scratched screens (that won't cost the earth)? What would be the best product to use, and procedure for polishing out a scratch?
Jason
Answer:
You can't do it. The anti-reflective coating will be damaged by any polishing
attempts, giving you a blemish that's worse than the scratch. All modern
monitors have such coatings.
You could try to minimise the visibility of the scratches with some clear wax that approaches the reflective index of glass. Car windscreen repair goop might work, provided it doesn't give you a big splodge on the screen around the scratches.
Lots of people have tried these sorts of stunts, though, and the general consensus is that once your screen's scratched, that's it. You're boned. It's like scratches on your sunglasses.
If you can't stand the scratches, then retire that monitor to server or game-box-for-visitors use, and get a new one.
And tape cardboard to the front of it the next time you take it to a party!
And on it goes
Your recent letter [the last one in this column] reminded me about something I saw on Today Tonight a while ago. The story is here.
I've had a quick look at the FuelStar site and it seems to be on the level. I was wondering if you could cast your well trained eyes over it.
For people like me it's more tempting as it would let my old leaded engine run on unleaded fuel as well as increase efficiency. Double savings :-)!
John
Answer:
Wow - a Story No Motorist Should
Miss!
The discovery that homeopathically small amounts of tin added to the fuel stream yield magical improvements is, if you ask me, questionable.
It's the usual deal. Amazing claims, with supporting pseudo-evidence from weird unverifiable places. The Ministry of Science and Technology, Thailand, eh? Plus somewhere in France, and some outfit with a FrontPage template site that gives no real idea of who the hell they are. Plus a smattering of mechanics and journalists, of much the same consistency as the people you can get to attest to the reality of a perpetual motion machine.
And that's about it for supporting evidence for an invention which, if real, would be worth many billions of dollars. Not even a freaking university on the list; how hard would it be for them to hand a hundred bucks each to five engineering students and get after-hours testing done on a bench engine at a technical college somewhere?
They make the claim; they get to prove it. We're waiting. I'll bet you a dollar that we'll wait forever.
As for letting your old engine run on unleaded fuel without modifications - if it does it with the FuelStar, then it'll do it without. The tetra-ethyl-lead in leaded petrol is primarily there as an octane booster, not as a valve seat protectant. Most engines made to run on leaded petrol will also run fine, without undue wear, on unleaded petrol with the same octane rating.
