Slashdot Mirror
Truth in Advertising?
PerformanceEng wonders: "I work as an engineer for a large technology company in the U.S., and have been privy to what I find a interesting practice. It's well known that marketing data sheets often paint the best picture of a product while leaving the devil in the details. I've come to expect this, and when I am evaluating technology, I always have a skeptic's eye for claims made by the sales and marketing folks.
However, I've also witnessed our product go into test labs (usually for the purposes of running a series of tests for a 'bake off' in a trade publication). Not uncommon is the attempt to 'tune' the configuration of the device under test to perform in the best light (not unlike tuning your car to pass emissions tests). I have seen it go as far as exploiting weaknesses in the test that, if the test operator discovered, would be considered bad faith. To the other engineers: Are you aware of this kind of practice at your company? To the IT professionals: How much faith do you put in these sorts of publications and their 'bake offs'? To everyone: When does spin doctoring cross the line and become false advertising?"
I've also witnessed our product go into test labs (usually for the purposes of running a series of tests for a 'bake off' in a trade publication). Not uncommon is the attempt to 'tune' the configuration of the device under test to perform in the best light (not unlike tuning your car to pass emissions tests). I have seen it go as far as exploiting weaknesses in the test that, if the test operator discovered, would be considered bad faith.
:-) Seriously though, this has been the whole problem with "benchmarks" like SPEC and others that ultimately results in pissing matches between manufacturers saying "my product is faster than yours" which for 99% of the users out there means nothing. In fact, even for that 1% of us where it does make a difference, specific optimizations to ones code or algorithms typically will get you more performance. So, what it really comes down to is how productive is the product + environment + task that you are assigning to the platform.
Oh, you work for Intel then.
To answer your question of false advertising, I would say keep to the standard that most of us scientists do: Specifically, peer review and ensure that your results can be duplicated by said peers. If results cannot be duplicated, then it is false advertising.
Visit Jonesblog and say hello.
As a person who worked in the advertising business side, I can say wholeheartedly that truth in advertising is a complete misnomer. The whole concept of advertising rejects the idea of truth. I don't sound bitter do I?
Stay tuned for new sig...
This isn't just a phenomenon in the IT arena. Have a look at medical journals some time... You have to be VERY careful when putting stock in the findings of studies -- the first thing to check is who *funded* the study.
I think it's just a fact of life: everybody wants their product to be seen in the best light, and to sell well (in the case of commodities or services).
That's why Amazon.com has reader reviews, sites like epinions.com exist, and Slashdot has moderator points. It's also why there are hardware review sites -- we can't just trust the manufacturer's PR now, can we?
So, people may be inherentely biased and often untruthful, but with proper monitoring (read: community involvement), the truth will out.
Marketing materials do not set out the faults of the product. This is not the role of marketing. Marketing aims to connect buyers to sellers. Providing information about faults does not help to make that connection. Also, many of the "tests" cited by marketers are labeled with titles such as, "Customer Success Story". This should be a clue that the material will not detail unsuccessful characteristics of the product.
Finally, marketers in most companies are not technical experts. They have to rely on the information provided by engineers and programmers. Many companies avoid ever telling the marketing department anything negative. As a result, in many cases, marketers aren't lying when they make claims -- they're explaining what they were told. Many of these marketers, especially the ones writing up collateral, are junior, new to the company, or even working on contract, so they don't have the depth of knowledge to tell that they've been given misleading information. Other people in the company sometimes lie to the marketers. It's not always black and white. (Not that all marketers tell the truth, of course.)
-- SYS 64738 --
Yes, I've seen this sort of thing at other places I work. It's inherently dishonest. It's justified via a) claiming that it'll help sales (dubious), and b) claiming that everybody knows that they're bullshit anyways. Note that the two justifications are mutually exclusive. Doesn't stop them from using them though.
No, I trust none of these "bakeoffs". Or any other IT advertising for that matter. There isn't a single mainstream IT rag which is even marginally trustworthy. Go ahead and, instead of reading just the bakeoff that you're looking for, read an article about something you already know about (through hands-on experience with all the primary alternatives, including a FOSS alternative if it's software and there is a FOSS alternative). Note how much stuff they get wrong, how shallow the article is, and how it almost reads like an advertisement. The same is true for cars too, largely, at least from what I've read. I can't comment on other industries since I'm not particularly familiar with their trade press. Note, however, that I still don't trust them at all - I expect they're just as bad. It's just that I don't make enough decisions relating to those industries' products to warrant reading the trade press - instead I go to the store and carefully examine the alternatives.
This sort of thing crossed the line into fake advertising at least a decade ago. Companies routinely make absurd claims and get away with it. There's just no political interest in enforcing it. At best they'll include fine print in their ad. If it's a print ad, maybe you'll be able to read it. It's been a while since I've seen an ad with fine print whose fine print didn't take up at least 10 lines of extremely small type. Television ads are a joke, it's impossible to read the fine print at broadcast resolution, regardless of the size of your TV, and it typically takes up a whole screen.
What can we do about it? Elect governments with some spine. These sorts of advertisements will continue to be successful so long as people are poorly-educated, and people will continue to be poorly-educated unless there is a strong collective agreement in place that says "yes, everybody needs some minimum level of education, otherwise they're prone to manipulation and our society is controlled by those who control the media or the other forms of information dissemination." It's funny, isn't it, how political campaigns in the US almost exclusively take the form of commercials? (Except for the "debates", which are a joke to everybody outside the country.)
Note that when the US was founded, everybody who advocated democracy made sure to point out that the requirements for democracy included an educated public, free speech, and free press. People have totally forgotten the education bit and the press bit. (A government-controlled press is no more effective at disseminating important information than a press controlled by an aristocracy - corporate or otherwise.)
Barclay family motto:
Aut agere aut mori.
(Either action or death.)
Hell, not just drivers... BIOS!
In the mid-1990s my company did quite a bit of graphics card testing (still does, but it was much higher profile back then.) It was pretty routine for us to get baked drivers (and there were some very impressive cheats) less routine, but still common was to get a board with a BIOS cheat, which would do anything from altering its own board timings to be out of spec (sort of "overclocked out of the box") to running code that would adjust the PC's heartbeat interrupt to slow the clock ticks to make the board appear faster if benchmarked using the PC's own clock.
In the end the best solution we came up with -- because we worked with a lot of alpha/beta silicon since we tracked chips more than boards -- was to more or less formalize the cheats and what was/wasn't permitted, and also to give the companies that submitted alpha/beta hardware to pull the results before publication, so that if one company pulled a fast one, the others that would be look bad in comparison simply wouldn't be compared; this resulted in a sort of a stalemate of cheating.
The most extreme (but permitted) cheat I ever encountered had the company involved paying over $100,000 to have a custom graphics driver written overnight that incorporated an optimized version of parts of the DirectX rendering engine (this was ~ DX5 era). When they found out their primary competitors pulled their boards from testing, you can imagine they were less than pleased.
The point of all this: a competent testing lab, particularly part of a magazine "shootout," should be well aware that cheating is taking place, and prepared to identify major cheats. Back in the heyday of PC Magazine in the mid-90s, their benchmark people were top notch and the benchmark ran a considerable number of cheating tests to clear out the more bogus attempts.
Oh, and you can be pretty assurred your competitors are doing the same things you are.
Based entirely on your comments, I would suggest that is the true strength of Consumer Reports' reviews--you have not just a ranking, but also a detailed explanation of how that ranking was arrived at.
The people who buy based only on a final arbitrary score or ranking are just as screwed as the people who choose a CPU based solely on its clock speed, or an audio amplifier based solely on its output power. Sure, such people exist, but there's useful content in CR for those who are willing to look.
One hopes that people willing to plunk down the cash for a copy of CR are also willing to spend a small amount of time reading the whole article before they buy a twenty thousand dollar vehicle....
~Idarubicin
Wait, you're responsible for getting unstable drivers onto the WHQL? I'll kill you. I've had it up to here with unstable drivers. I lost a RAID set to unstable drivers that passed the WHQL.
I mean, cheating on benchmarks is one thing (the card is just slower than it benchmarks), but the WHQL is supposed to be a stop gap measure: sure, it's Windows, we hate Windows, whatever. But where I work, we use it. And WHQL drivers are something that you're supposed to be able to lean on; they're drivers that may not be the latest, greatest, but they will work.
I can't tell you how long it took to track down that the RAID WHQL certified drivers were the problem. It's something you're supposed to be able to put a little checkmark next to when diagnosing problems, a "it can't be that!".
I am John Hurt.