Slashdot Mirror
Domestic Appliances Guzzle Far More Energy Than Advertised, Says EU Survey (theguardian.com)
Chrisq writes: An EU study has found that many electronic devices and appliances use more energy in real-world conditions than in the standard EU tests. Often the real world figures are double those in the ratings. Sometimes this is achieved by having various optional features switched off during the test. For example, switching on modern TV features such as "ultra-high definition" and "high-dynamic range" in real-world test cycles boosted energy use in four out of seven televisions surveyed -- one by more than 100%. However some appliances appear to have "defeat devices" built in, with some Samsung TVs appearing to recognize the standard testing clip: "The Swedish Energy Agency's Testlab has come across televisions that clearly recognize the standard film (IEC) used for testing," says the letter, which the Guardian has seen. "These displays immediately lower their energy use by adjusting the brightness of the display when the standard film is being run. This is a way of avoiding the market surveillance authorities and should be addressed by the commission."
Very Widespread
People game standardized tests. Graphics cards, benchmarks, cars, students, teachers, if you have a standardized test, people will put in the effort to game the numbers.
Maybe they should do what they do for TV : recruit a random sample of people, stick an energy monitor on their appliances, and see what happens.
Ah, right, so it's not that companies will deliberately try whatever it takes to lie on tests, it's that the tests suck.
Next they're going to tell us that automakers somehow game the emissions tests. Yeah, like THAT'S possible.
-Styopa
The EU/US should just buy a kill-a-watt or something similiar, it's really just a glorified multimeter that does some of the math for you.
Then do what Consumer Reports does, buy a few units of each model at random stores - run all the various models for several weeks continuously (if a fridge, if a TV, have a 8h daily period) under a standard protocol, not test, and the meter should output a reliable weekly, monthly, and yearly usage. I do this at home and am able to narrow in on my actual electrical usage. There's little fudging to be had if the usage is real versus some standardized testing. It's akin to making every student in class read a random section of the same book out loud in class (and changing the book every 6 months) rather than giving them the same standardized multiple choice test year in and year out to gauge their literacy.
Just make it law that the manufacturers have to reimburse the EU/States for 5 units of each model product it wants to sell there annually, otherwise they aren't allowed to sell their wares. Testing will be randomly dated.
My boss slaps a folded-over InfoWorld magazine onto my desk, thick enough to kill a rat with in those days. He says with obvious glee, "How bout dem apples?" It is Steve Gibson's INFOWORLD column of March 8 and Gibson (with obvious glee) has caught a manufacturer of Hercules graphics cards red-handed. The standard WinBench program had conducted a series of tests --- and in one particular test of text display, in which the phrase "The quick brown fox jumped over the lazy dog's back then sat on a tack" is continuously painted on the screen --- the card performed oddly spectacularly. It was that one score that when combined with the others, ranked the card above the competition. Suspicious, Gibson changed a single letter in the test phrase and the card's score dropped to a reasonable range. The card was apparently recognizing that a test was in progress and 'cheating' by failing to actually over-write this static text repeatedly.
I love the comment by the manufacturer when Gibson contacted them (read it!) but what intrigued the industry the most was that the cheat was not to be found in the Windows driver code, it had been embedded into the firmware of the accelerator chip. In the next Winbench version the test phrase jumped around the lazy screen's back during the test, rendering the cheat obsolete.
Has anyone done an energy study to estimate how much energy is consumed by EU "market surveillance authorities" and even the EU apparatus itself? Perhaps if we recognize the EU as a special case and stub the whole thing out with a rubber stamp, people will be able to watch HD television and toast four slices of bread at once and with former EU personnel in the workplace everyone will be able to work one less day a week with same pay.
<blink>down the rabbit hole</blink>
Well, in a sense, the testing procedure sucks for not accounting for different modes of the devices and not mandating a worst-case-scenario or at least a middle-of-the-road mode to be used during testing. Or mandate the use of the modes the applicance will have when taken out of the box and or the ones it recommends to users. For example, most TVs have a power-saver mode with the backlight set very low, but out of the box they default to "standard" or "dynamic".
And increase the cost of getting the testing done, and therefore increase the cost of the product. Our washing machine started giving us issues, and since we reuse the gray water coming out of it (watering plants, but mostly for flushing the toilet) so we had a pretty good idea of how much water it used. So with the machine giving us issues we were in the market for a new washing machine. Shopped around and bought an expensive Bosch washing machine with awesome ratings for energy and water usage. Turns out it uses MORE water than the old standard one. Not sure about the energy claims because that is one thing we have not been keeping track of.
There are three kinds of falsehood: the first is a 'fib,' the second is a downright lie, and the third is statistics.
See: Goodhart's Law, "When a measure becomes a target, it ceases to be a good measure".
In this case it was obvious a long time before the VW thing that the power consumption figures were understated, although the specific usage of defeat devices of the kinds being described is new. Anyone who has deployed some of those power monitors that sit between an appliance and the socket to see how much power a given device is drawing over time will be well aware that peak, average and idle power draws for a given device are typically above those stated, and often by a considerable margin. The only real question here is why it took so long for those that were doing the regulatory tests to realise that something was amiss and dig a little deeper - more average Joes complaining after deploying smart meters in efforts to go green, perhaps?
UNIX? They're not even circumcised! Savages!
If you have standardized tests, people will do what's necessary to perform well on those standardized tests and ignore anything else. What happened when schools got them? Every teacher began teaching to the test, i.e. what will be asked at this test, everything else was simply swept under the rug. Why? Because it won't be tested, so it's superfluous. Actually harmful, because it will take up valuable time and brain capacity for no gain.
No gain at the test, that is.
Same here. Your test will perform X, so we'll do good at X. And on nothing else.
There's also that problem that customers want cheap TVs that have great features, and that is pretty much the exact opposite of power conservation. You cannot build cheap TVs that have all sorts of features, great resolution, high contrast, fast switching and so on, and don't consume much power.
Now take a wild guess which of the three things "cheap", "performance" and "compliance" gets thrown out the window? Hint: You can't fire cheap, because that's what both the maker and the customer wants. You can't cut performance, because the user would eventually notice and a huge stink ensues on various test sites on the internet. And compliance is something that gets tested once and nobody really gives a fuck about it.
So pick the one that you could do without.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
If the test can be gamed, it sucks. Period.
What you say here is the equivalent of Software having bugs isn't the problem, it's all the malware's fault if you get infected. Yes, the malware abuses those faults, but the faults enable it in the first place.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
If you have a test X, we will do good at X.
If you don't have a test X, then we don't know if you will do good at anything.
The only people that gripe about tests are the ones that haven't learned anything and don't want to get caught.
The article is indeed worth a read. I couldn't imagine a maker of hardware saying today "Yeah, I wrote that cheating routine myself, that way we come out on top every time a comparison is run, pretty clever, eh?"
They may THINK that, but I doubt they'd have the chutzpah to just throw that in your face.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
The people at Phoronix should take heed to these words; but they are very good, they probably already know these tricks... I was thinking how much chaos such things cause on user forums.
As to the practice being "very widespread", one at first would think this is Capitalism working and actors in that system actively looking for a competitive advantage. Except it is not so when you subvert the rules. Oh, it's clearly advantageous but it is sidestepping competition -- thus not Capitalism as we know it, because some would avoid the market regulation that promotes the best product (quality/cost) for consumers.
Much like China is communist only in certain aspects, but a savage capitalist in economic terms, corporations are seeking to be immune to competition. As a consumer, it tells me to also get out of the game. Many decades ago a teacher I had said commercial shower labels were mostly factually wrong: makers would advertise low power, yet a shower would heat water to a quite high temperature. That effectively render labels useless. And one has to rely on benchmark sites (like Phoronix for IT) and hope they can circumvent these lowly tricks. Of course, just using such trickery is enough for a company to descend into the abyss of untrusted brands.
Which is only fair, because it allows entrants who play by the rules to get promoted. Suddenly you may want that Chinese "copy" if you know it is an honest product.
Yet somehow in the good old USA, we seem to think that the EnergyStar rating is this game-proof metric, and SlashDot blows a fuse when Trump issued a recommendation a couple months ago to shut the program down. It's almost like he recognized the whole program is rife with fraud and should be shut down.
Testing is necessary, but don't test subset A and tell everyone up front that you'll test subset A. Same as in school, what did you learn when your teacher told you that you're going to get tested about the stuff on pages 80-110? You learned the stuff on pages 80-110. If it was on page 79 or 111, it didn't even cross your mind to learn any of that. Because it would literally be useless knowledge.
Should you have learned it? Yes, of course. Because pages 60-80 explained just what 80-110 required you to know to understand it. But you didn't, you learned 80-110, didn't understand it and just crammed it into your head for the test, swiftly followed by blissful forgetting right after the test was over.
With testing like this, it's no wonder that we don't get anywhere. In school and technology. We have to make sure that tests cover the whole spectrum of what's required. And if it's impossible to test it all for some reason, make sure that your test subject doesn't know what part you'll want from it and thus HAS to be prepared for everything.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Increasing the cost of bringing an appliance to market might not be too bad a thing. Have you tried to read reviews for white goods recently? The manufacturers churn models so quickly that by the time one has been reviewed it's no longer being produced and so you have to just hope that the next model has similar characteristics. Giving manufacturers an incentive to keep them on the market for a bit longer would be beneficial to consumers.
I am TheRaven on Soylent News
After the shitstorm VW got, it should've been obvious to other companies that this sort of BS really doesn't pay in the long run. Unfortunately, that doesn't seem to be the case.
--
So you spent more on the one that is apparently gaming standardized tests...? I'll just point out that the work to game the test isn't free, either.
toast four slices of bread at once
*sigh* These Euro-Myths never die, do they? No matter how often they are debunked, they just keep coming back.
Here is the source of the claim, it's literally one sentence on page 56: http://www.ecodesign-wp3.eu/si...
A bunch of liars, sorry "journalists", claimed that this meant the EU was going to ban two slot toasters. Such a plan never existed.
Later a new variation on the claim referred to 4 slot toasters because the EU was considering minimum efficiency standards for heating and cold storage kitchen appliances. Of course, there was never a ban - you can make a 40 slot toaster if you want, it just has to use reasonably efficient heating elements and mechanical design.
watch HD television
We have had HD television broadcast over the air for more than a decade in Europe, and you can't buy new SD televisions any more.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
While designing my solar system, I spent a few months logging data with my kill-a-watt. Not only did I learn how much energy things used, I learned how to vary loads to maximize daytime energy use, greatly reducing the amount of storage (batteries) I needed for night.
"You learned the stuff on pages 80-111"
Yes, at least you learned it. The problem with not testing is you don't know if they learned pages 80-111, or learned anything at all.
"We have to make sure that tests cover the whole spectrum of what's required"
So you want testing, or you don't want testing? People complain about Common Core testing all the time in schools and "teaching to the test". But they never offer any alternatives other than "make sure everyone learns everything"
If its in on the budget, fine. It is the new age of coal energy that would pose the difficulty for the Modern Man.
...who really uses the "green" program on washing machines or dishwashers? If the artifact mostly soaks in lukewarm water for 4 hours and comes out still dirty and with remains of detergent, it has to be washed again. No energy was saved. One can always set goals, but even bureaucrats cannot bend the rules of physics just by creating arbitrary standards.
Has anyone done an energy study to estimate how much energy is consumed by EU "market surveillance authorities" and even the EU apparatus itself? Perhaps if we recognize the EU as a special case and stub the whole thing out with a rubber stamp, people will be able to watch HD television and toast four slices of bread at once and with former EU personnel in the workplace everyone will be able to work one less day a week with same pay.
I can only speak for lighting appliances here, it might be different for other books of the ecodesign guidelines.
Market surveillance authorities are the responsibility of the nations and usually completely underfunded to even begin to fulfil their purpose. Usually participants of the market (manufacturer, reseller, etc.) check on products of their competitors, because they can force injunctions against their competitors if products are wrongly labelled and push them this way out of the market.
Beside that we have a couple of NGOs (e.g. Stiftung Warentest) which check regularly on products and finance themselves by publishing the detailed results.
All in all its quite effective for the money spent, even if it is not by the letter or intent of the law.
The press likes to lick the pompousness of the supporters of the various anti-EU populist parties to get the copies out and internet ad space sold. By lying, if necessary. Meanwhile their production processes and journalists benefit from the standards and the freedom of movement.
There are different ways of gaming tests. For example, if a CPU manufacturer knows that a particular test suite is likely to be used which does a lot of integer multiplications and not so many subtractions, they might optimize their processor to be faster with those instructions at the expense of others, resulting in lower real world performance but higher performance on the test. That would be a problem with the test.
However, actively recognizing a particular test film and then changing settings to lower image quality and energy consumption for that particular film, that's a whole different story. You can't blame the test for that, it's simply deliberate fraud.
What else are the testers to do? Show random films? Then the manufacturers will complain that the test for their TV set had more bright scenes and therefore was unfair to them. Tests have to be identical, so there's no way around using a standard film.
When you give everyone the exact same test over and over again don't complain when people learn the answer. Yes the test sucks. It's the lazy way to do it.
Seven puppies were harmed during the making of this post.
"Lightbulb lasts for 5* years"
*Testing was done by turning it on every other season after forgetting about it after a year because we put it into the attic closet
The issue is the nature of the questions. While there's absolutely a place for multi-choice, or questions that only have one correct written answer, the only way to test on whether a student has truly comprehended something is to get them to explain it, or better still extrapolate from it, in the test. The problem with that approach is that it introduces more ambiguity into the scoring and actually requires that the person (and it currently has to be a person) marking the test knows something about the subject(s) at hand. Unfortunately, that means a lot more expense than simple machine readable multi-choice forms or easily outsourced reviews of basic written questions, which isn't exactly compatible with either for-profit education systems or those that have tight budgets
UNIX? They're not even circumcised! Savages!
Make the standard file "Star Wars," or something else that would enrage the mobs if the film quality were poor. Or a suite of several movies. After all, testing isn't a 15 minute affair. Let the films play back to back overnight while equipment is gathering results.
--Jim (me)
I don't typically walk around with appliances, I tend to set them up somewhere in my house and use them in place. I find it a wee bit onerous to walk around with my fridge.
Try turning off the "dynamic" option.
Graphics drivers were (and maybe still are?) notorious for detecting benchmark software
Oh they most certainly still do. But the plus side of it is, they've taken this technology to detect benchmarking software and expanded it to detect what game you're playing and load a custom optimization for popular games as well. So it's not all bad.
Not likely. Smart meters are one of those things forced upon consumers by power companies whenever they get an alternative power source like solar panels installed.
In California, they tried to force them on everyone, period. But they are shit. When they fail they fail in favor of the power company, or they let smoke and fire out. Meters that have been working for literally 30 or 40 years get replaced with "smart" meters that fail in 30 or 40 days. Not every time, of course, but way more than is acceptable. PG&E even gave a third party contractor the gate code so that they could come install a smart meter after we had formally opted out. PG&E is literally evil in every way in which it is possible — they do, after all, willfully kill people for profit, so this negligence is just the tip of a very large iceberg... which is melting.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
But this is amazing! I had no knowledge of the EU myth, sole inspiration was my own four slice toaster (manufactured ~1993, Chicago) that has a switch.
Most four slice toasters have two levers, each of which causes two slices of toast to sink into the machine and be toasted separately, as they are actually a pair of two-slice toasters in one case. This has long been true.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
A month after I put my common area lights on SmartThings I saved $10. After putting most of the rest of my house on SmartThings light switches, I am now saving $30 per month on my electric bill. I am now able to see and turn off lights that my kids habitually left on when I am not home. I would say many consumers are not being "forced" to put in monitoring but want to see exactly where their money is going.
Tests have to be identical, so there's no way around using a standard film.
The obvious answer is to keep the test corpus a secret, so that people can't design to the test. Then you only need some mathematical way of proving that the test is meaningful without revealing the footage.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Just dishonest companies. The difference between the EU and the US, the EU has rules against this.You want to game the system in Europe? Sure, go ahead. It's a self-certification. You can game it all you want, you probably will get products through, no problem. But wait until you get caught because someone reported you and questioned your documentation, then your top level employees can't even get into Europe because they'll all have an arrest warrant for them. Then you can kiss all your prospects in the EU goodbye.
I've been watching the EU clamp down big time on this sort of thing in the past two years, especially radio interference tests which is required for all electrical products. Many companies are completely ignoring these required tests and I've seen many burned, especially in the US and China (You think the US doesn't try to ignore things like China? Sometimes they're worse because they argue why they have to do it!!).
Already watched plenty of companies burned for thinking they can just put the CE mark on everything they want only to get burned by their own client when they questioned their documentation of use of standards and don't comply with anything.
or you could switch to LED bulbs
love is just extroverted narcissism
Putting together rules on optional functionality is also a minefield, should a TV that uses 50% less power than a rival in normal mode be penalized because it can offer HDR (the rival can't) but when doing so it uses 10% more power
Marketing gives the answer.
The energy usage when used with HDR needs to be as prominently displayed compared to the energy usage when used without it as the marketing material is for highlighting HDR compared to non-HDR.
If they market it as a HDR TV, that use is what needs to be highlighted. If they also in smaller letters say what the lower energy usage is if not using HDR, that's fine.
Just like a car marketed as a city car should highlight the city mileage, and a truck marketed for hauling heavy loads should be mileage tested with heavy loads.
Students caught cheating fail their exam. So these TV sets should simply be taken off the market.
vw was just the highest profile case to be publicized. companies have been doing this for decades, at least. probably since these types of tests, or benchmarks, etc. started.
in the case of televisions, samsung, lg, and others have already been caught cheating on standard efficiency tests in europe and the u.s. as profiled in mainstream news and media since mid last year or earlier.
This is no different than a certain car maker's vehicles knowing when they were being tested and responding accordingly or how a certain graphics card manufacturer built drivers that would know when they were being benchmarked and adjust the behavior of the card.
LK
"Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano
I see. So you don't mind testing, but dislike multiple choice questions? Not all Common Core tests are multiple choice. There is essay writing for example. But yeah, keep moving the goalposts. There is nothing wrong with standardized tests. They aren't the end all be all, but there needs to be measurement - even if it has flaws.
They do, it has been long known by anyone over 30, oh millennial. They also continue to use power if they are turned off but plugged in, physically unplugging unused appliances will noticably reduce an electricity bill. It's one reason rechargeables are not a panacea, one must still charge them. Again, this is common knowledge to anyone born before 1978. I think we were actually in better shape in terms of understanding these things 30 years ago, the seeming magic of 21st century tech makes people think it actually IS magic. Nope. It's the same as its ever been, it's just more well-hidden from the user now. Kvetch and pontificate all you like, but it's true.
Did you even RTFA? They are blaming the smart meters for the high readings, not the bulbs!
love is just extroverted narcissism
They made it look as though this was the first ever violation of any kind by a car manufacturer and somehow more evil than anything else that had ever happened, even though it quickly became clear how widespread similar tricks are.
So, they cheated, they lied about how they cheated, and they became the world's largest car manufacturer as a direct result of the fact that they cheated, but the anonymous cowards are popping up on slashdot saying it's all political.
"Everybody cheats, why single out VW merely because they did it on a larger scale and deliberately" is not an excuse.
No - have a "test film of the month". too little time to add "recognition", and still fair. All tested in the same month get the same movie. More fun for the testers too. . .
Since in any given month only one new television model is likely to come out, in practice that would mean every television would be tested using a different film.
If different films require different amounts of energy use, that would lead to a test that is randomly harder to pass for some models and easier for others.
What the fuck does a single video clip have to do with standardization? Nothing. I've taken standardized tests before. Like the USMLE. Guess what. There is more than one question and the questions are designed in such a way that just memorizing stuff will pretty much guarantee failure. But of course a lot of time goes into question design. Playing a single video clip and attaching an oscilloscope to a TV is a really lazy way to do things.
If you're making a test, you expect cheating. Why? Because if everyone was honest you wouldn't need a test in the first place, you could just tell manufacturers to comply and expect them to do it. Therefore if you're designing a test to ensure compliance, you need to design it to catch cheaters not encourage them.
Seven puppies were harmed during the making of this post.
Not true. And VW did not have "some of the lowest real-world NOx emissions."
It is true that, after the VW scandal, investigations revealed that six other car manufacturers used strategies in optimizing their emission controls to lower emission in testing but not in real world conditions. But VW did not merely optimize their controls for test conditions-- they actually had software to detect the fact that testing was going on and turn on emissions controls that were off the rest of the time.
VW deliberately, consciously cheated, in order to make the claim that diesel was "clean" and take over the car market. (And to avoid paying for the Mercedes emissions-control technology). They cheated more, they cheated worse, they cheated more flagrantly, they cheated deliberately.
"Selective enforcement protects General Motors' market position."
What could be wrong about that?
Your 'direct result' assertion is unadulterated bullshit. Insert quarter to try again.
Usually I only toast one slice of bread-- but my toaster heats up both slots anyway. Why don't they make a switch to allow me to toast just on one side?
Your 'direct result' assertion is unadulterated bullshit. Insert quarter to try again.
Classic diesels got better gas mileage but had worse emissions. Volkswagen made the claim that they had solved that problem: they could make diesels get the better gas mileage and also get low emissions... and also sell at a reasonable price!
They were partly right. When they geared up to break into the US market with their diesel passenger cars, there actually was a pretty good low-emissions diesel technology... but Mercedes owned it. Their original plan was to license the Mercedes technology, but the company had a change in CEO, and the new CEO decreed no, we won't buy another company's tech, we can develop our own.
But their home-developed tech wasn't harder than they thought, and it couldn't simultaneously meet the emissions standards, still get the gas mileage, and have performance acceptable to Americans. Unless they cheated.
So, the bottom line: they broke into the US market with their "clean diesel" for exactly one reason: they cheated.
They didn't do it on a larger scale, aside from being plain bigger.
Wrong. They did it on a massive scale.
We found many car manufacturers whose cars were tuned in some specific way to beat the regulatory tests, and performed horribly in any real-world scenario--typically on-par with the VW offerings.
You missed the point. VW did their cheating not by merely choosing settings that performed well on the test but not as well in the real world. VW actually cheated: they detected the test, and turned off their emissions controls.
Yes, other companies also had poorer performance in the real world. Check for example, the Guardian article: https://www.theguardian.com/en... ."
"the diesel cars [from other manufactures] passed the EU’s official lab-based regulatory test (called NEDC), but the test has failed to cut air pollution as governments intended because carmakers designed vehicles that perform better in the lab than on the road. There is no evidence of illegal activity, such as the “defeat devices” used by Volkswagen
Test films don't have to be identical, they just have to be equal. Chopping up the video and mashing it back together in a random order would keep things fair, while making it harder to detect. There are other tricks that could be done to make sure that the test videos are all equal, without being the same. Also they would be run multiple times and averaged (which I believe is already the standard procedure) so that small differences induced by sudden brightness changes would tend to even out. (It's not like real films go out of their way to avoid such changes.)
How is the Riemann zeta function like Trump rallies? Both have an endless number of trivial zeros.
The question should be, in the absence of cheating, would the tests reasonably reflect real-world usage?
If the answer is no, then the tests need to be updated to better reflect real-world usage.
If the answer is yes, then there is no problem with the tests themselves, and it's the cheating that needs to be addressed.
You could attempt to address cheating with more comprehensive testing, but it would likely have to be a *lot* more comprehensive, since anything less than "the entirety of normal usage" is a subset that will inevitably be optimized for at the expense of the rest. You're basically trying to find a technical solution to a political problem. It's not impossible, but it's probably one of the more difficult approaches.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
There is at least one very good reason to recognize test conditions: predictability of test results.
As a company, you perform in-house testing to understand the characteristics of a device prior to sending it out for official review. You don't want any surprises. The test conditions are public, and known (as they should be). So, rather than rely on the competence of the official testers (or lack thereof), you make your device recognize the test conditions, and put it into a standard configuration. That way, if the testers were playing with the brightness, loudness, color balance, whatever, to examine the item prior to testing, your product won't get an erroneous assessment because they forgot to reset it from, for example, eye-bleed-level back to normal brightness. It makes sense.
Same goes for the VW case, and for the cases from the other auto manufacturers: the test conditions are most certainly not a standard use-case and must --- in the case of automobiles -- be recognized as such to avoid treating the highly anomalous conditions as an emergency situation, e.g. front wheels at driving power while rear wheels are not rotating. Again, perforce the item must recognize a test to reset to a standard configuration in order to ensure predictable, repeatable testing results.
Now, if you accept this premise that there should be a standard configuration for testing, and the device (be it car, TV, whathaveyou) should be able to recognize the testing conditions and reset to that standard configuration, we can now take a step forward. Cheating is not resetting to that standard condition. Cheating is selecting a standard condition that deviates significantly from either default conditions, or from normal operating conditions as selected by real-life users.
So, resetting to a baseline is not prima facia evidence of cheating. What's really happening here, in this case? From the summary, it sounds like perhaps commonly used modes are not the ones selected during the testing, which might be evidence of cheating.
Put my fist through my alarm clock with its ding-dong death inside my ear. - The Blackjacks.
Perhaps one film with the scenes randomly reorganized?
--- Most topics have many sides worth arguing, allow me to take one opposite you.
The obvious answer is to keep the test corpus a secret
Security by obscurity would provide an incentive for corrupt government employees to either leak the test details, or cheat by misreporting the results, since independent verification would be impossible.
In addition to helping people buy computers, I also help people buy TVs. A lot of them complain that the TV doesn't seem as bright as it seemed at the store, and will "randomly" suddenly turn very dark or off. That's my cue to visit their home, go through the TV's settings menu, and shut off all the power-saving features like the auto-dim timer and dynamic brightness.
So I end up visiting their home, turning both features off, raise the backlight brightness level a hair (not torch mode like at the store, but the proper brightness for the room), tweak the contrast and tone down the sharpness so there's less haloing. The owner thinks I'm a miracle worker and thanks me profusely for "fixing" their TV.
People want bright images on their TV. If you attempt to save power by too-aggressively mandating that TVs have "features" which automatically dim their image, people will just turn the features off. Real-life use does not deviate from the test because of manufacturer malfeasance. It deviates because the test fails to accurately model people's real-life behavior.
The only solution is strong consumer laws. In the UK products must last a "reasonable length of time", which in practice means if things like white goods that you would expect to last you a decade fail after three years you can get at least half your money back. The exact amount can ultimately be determined by a court, but generally if it lasts half as long as you would reasonably expect you get half your money back or a warranty repair.
This creates a great incentive for shops to sell good brands that last, because it's the shop that is on the hook for the cost.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
But it is not fraud, it is the manufacturer setting up the TV to the most conservative, test-favorable settings while the test is being run (this is completely reasonable and expected). It is possible that the TV can be viewed in this power saving, low resolution state, but if we the viewers want that UHD with HDR and high brightness, the TV can use more energy to accomplish this. The real culprit here is incompetent bureaucrats (surprise, surprise) and the test for not specifying a baseline resolution, brightness (in nit), and contrast ratio that should be adhered to for the test. Modifying the test parameters to define those settings and validating them with external sensors would create a consistent baseline for the test rather than wasting man hours coding firmware to recognize the test film.
This is akin to loading your truck to it's weight limit, doing jackrabbit starts from every stoplight, pumping the gas and brake incessantly while driving and then complaining that your car gets 2x the advertised mileage. The reality is that how you use your things affects how efficient they are.
If you disagree, please post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like
It wasn't so clear cut back then. People wanted to play Quake with the best possible frame rate, so the driver would detect quake.exe and apply some application specific optimizations. That was happening well into the 2000s at least, and was advertised as a feature by AMD and Nvidia.
All that has really changed now is that instead of the driver doing the tweaks, the game developers build them into their code. Somehow it's not cheating if the developer tweaks the pixel shaders when the game detects an Nvidia GPU, but it is if the Nvidia driver does the same thing.
Reviewers moved away from benchmark apps and now mostly post in-game frame rates, so the "cheating" is all fine because optimizing games is legitimate. Phone manufacturers still cheat on the benchmarks, perhaps because there are no tools for running repeatable in-game tests and accurately measuring frame rates on Android/iOS.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
If the answer is no, then the tests need to be updated to better reflect real-world usage.
I'm having an awful damned hard time getting this through peoples's heads.
Support my political activism on Patreon.
Your test will perform X, so we'll do good at X. And on nothing else.
So you just need to make your test comprehensive. If the car emissions test had involved fitting measurement devices to real consumer's cars at random, the cheating would have failed.
The EU vacuum cleaner tests are a good example. They test on multiple surfaces with a very good approximation of house dust, so the cleaning ability, energy consumption and emissions are all measured accurately. The main criticism is that they only measure with empty dust bags/bins, but the next version of the test is going to fix that.
You cannot build cheap TVs that have all sorts of features, great resolution, high contrast, fast switching and so on, and don't consume much power.
History demonstrates that to be untrue. TVs have been steadily getting more and more efficient over the years. CRTs became LCDs, CCFL backlights became LED, the image processing chip lithography got smaller and more power efficient, even as the amount of work increased. Standby power decreased by orders of magnitude too, and there were even savings from including set top box functionality into smart TVs. Most of them use ARM processors, which have got much more efficient mainly thanks to phones and tablets.
All the while the image quality has been getting better too. Contrast improved a lot when the change from CCFL to LED was made, for example.
The purpose of these regulations is to make sure manufacturers don't do what happened with vacuum cleaners. Bigger, more powerful motors because consumers equate big motors with better cleaning. In fact most of them just produced more heat, while cleaning much worse than Japanese models that used 1/4th the power, because in Japan consumers were prioritizing good cleaning and low power consumption. So now the EU puts a star rating on vacuum cleaners to show how well they clean, while limiting the motor size so that the manufacturers actually have to innovate instead of just applying more and more suction until it rips your carpet up.
This is why we have brush bars now. Available for decades in Japan, but not in the EU because consumers only cared about MOAR WATTS.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
Or you could, you know, switch to LED bulbs and pay $30/year for all your lights left on 24/7...
If you disagree, please post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like
The Nvidia driver actually DOES that same thing you describe. Check the patch notes for the GeForce drivers, they quite specifically talk about optimizing for a bunch of specific games every time.
-=This sig has nothing to do with my comment. Move along now=-
One of the main problems is features based on models rather than features as an optional addon or software upgrade. I mean DLC sucks in games, but producing 2 different TVs with two different model numbers based on some software features, or the inclusion of a SCART port instead of something else, just doesn't make sense. I think you'll find the number of "different" model of any whitegood on the market is actually quite small with only cosmetic / minor tweak changes underneath that none the less get a completely different number.
Your hope that the next model has the same characteristics is not far from reality.
That's two different things you and the GP are talking about. The GP is talking about life on the shelf. You're talking about life in your home. The latter hasn't been a complaint in this story so far.
Actually the aspects of both of your comments are intertwined, you can actually repair appliances for quite a long time after they cease manufacturing because many of the changes that come out in the new model are purely cosmetic.
*sigh* These Euro-Myths never die, do they?
A bit further up was a post complaining about bendy cucumbers being banned. Disregarding the fact that cucumbers are normally straight, people can't even get the myths right anymore.
That's a new one to me. Food seems to be a favourite topic of these myths.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
Later a new variation on the claim referred to 4 slot toasters because the EU was considering minimum efficiency standards for heating and cold storage kitchen appliances. Of course, there was never a ban - you can make a 40 slot toaster if you want, it just has to use reasonably efficient heating elements and mechanical design.
The problem in Europe is not the toasters, but the toasting products. Because sliced bread is now made very tall, toasters have become taller to accommodate it. However other toasted products like (English) muffins or crumpets remain the same size or slightly smaller than they used to be. This means getting my crumpet out of a toaster involves angling it and using the cancel button to eject my crumpet at speed whilst calculating the parabolic arc to ensure that it lands on the plate on the breakfast bar and not on the floor of my kitchen. All the time trying not to get burned because metal toasters are fashionable.
As I'm constantly getting 3rd degree burns, I end up using the top down griller to toast crumpets, which I'm certain is less energy efficient than a toaster.
Calling someone a "hater" only means you can not rationally rebut their argument.
No, I said there is absolutely a place for multiple choice, and was talking about testing in general rather than SATs specifically - which I didn't mention at all. The trick is to find the right balance between establishing that the student *knows* the subject, for which multiple choice tests are fine, and that they *understand* the subject, for which free form text/essays/dissertations or verbal discussion is usually much better. The problem is that the costs and skills required for marking the assessment go up as you progress from multiple choice, leading to more of a bias towards multi-choice, which in turn leads to more bias towards teaching the exam rather than teaching the subject. The solution is have a better mix of question types - multi-choice, set answer, *and* essay - that is appropriate for the subject at hand, but someone is going to ultimately have to pay more for it and that's where the real problems are.
UNIX? They're not even circumcised! Savages!
There is usually a rack on the top for crumpets and the like. Your aren't supposed to put them in the slots. I'm not an expert though, and I didn't RTFM.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
They made it look as though this was the first ever violation of any kind by a car manufacturer and somehow more evil than anything else that had ever happened, even though it quickly became clear how widespread similar tricks are.
So, they cheated, they lied about how they cheated, and they became the world's largest car manufacturer as a direct result of the fact that they cheated, but the anonymous cowards are popping up on slashdot saying it's all political.
"Everybody cheats, why single out VW merely because they did it on a larger scale and deliberately" is not an excuse.
VW's problem is two fold.
1. They got caught.
2. The way they cheated.
Everyone, including the regulatory agencies know you'll never get laboratory figures out in the real world. If I want to know the MPG of the 2er I'm getting, I'll ask people who already have said 2er (online forums for everything means this has usually already been asked and answered). However the state of tune of the engine in the laboratory must be the same as the state of tune of the engine on the road. VW cheated by changing the tune of the engine when it detected test conditions.
Everyone plans for the test, everyone games the test, but that isn't cheating. VW basically used a different engine configuration for the test and that is a strict no-no. Because stages of turbo tuning can change engine characteristics significantly, this would be like Ford taking the 4-pot mustang to the tests and then using those figures for the V8. Beyond just being shady and dishonest, It could end up costing owners megaquid if it wasn't discovered this early because here in the UK, as in many other countries cars older than 3 years have to undergo yearly testing which included an emissions test and the car might not detected the test conditions when parked up at the MOT centre. If you fail the emissions test your car is illegal to drive making the car worthless to sell and expensive to fix.
Also such tests do need to be carried out under laboratory conditions because they are comparisons between different automobiles. So we need to control as many variables as possible. Yes this makes it easy to game, but that is an unfortunate consequence for the need for an accurate comparison between disparate cars.
Calling someone a "hater" only means you can not rationally rebut their argument.
This is pretty much what graphics drivers do today, too. To optimize for a game when they detect it as the foreground application. And I highly doubt that anyone considers this a bad or even illegal practice. That's also pretty much what I WANT the driver to do, to get out of the hardware that I have the maximum for the game I play.
The complaint here is that this was done to artificial tests that had zero benefit for actual, real-life, applications but was used to mislead people into thinking it had.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
What else are the testers to do? Show random films? Then the manufacturers will complain that the test for their TV set had more bright scenes and therefore was unfair to them. Tests have to be identical, so there's no way around using a standard film.
There are solutions to this. You can release a new film every year and report performance across the multiple years of testing. That way you'll be able to see that the TV released in 2016 consumes much more energy on the 2017 film. You'll also be able to see trends from one manufacturer to the next that can point to monkey business
Disregarding the fact that cucumbers are normally straight, people can't even get the myths right anymore.
Surely you mean "can't even get the myths straight anymore".
The way I see it, standardization of testing is both necessary and self-defeating. Ideally, we want to ensure that people become useful adults with some ability to contribute to society. We believe that having a well rounded education promotes this goal. We want to make sure we are getting value from the resources we expend into this effort.
In order to ensure we are getting value, we need some way to measure the results of education. The measurement has to be applied equally across the board, or else things like styling preference of the test scorer unfairly impact the results of the test taker. Multiple choice tests do a lot to eliminate bias from test scoring.
The problem that we get is that multiple choice is at best a proxy for understanding. We use the proxy because it makes the process so much easier and so much fairer. But it doesn't actually measure the thing that we want. It is measuring whether people can draw the right conclusions from a simplified set of possibilities. This almost never comes up in real life. There is a reason that for graduate level degrees, an oral dissertation complete with a question and answer section is necessary to receive a diploma. This way, in addition to demonstrated ability to do something useful in a field (develop new research or create a new work), evaluators can actually see how well the student understands what they are talking about. But this probably wouldn't work at the grade school level. It simply requires too many resources to be an option for every child, and few people expect children to be able to demonstrate the level of mastery that we actually want.
It turns out that making Key Performance Indicators for important societal factors is difficult. It doesn't seem like we've come across the "best" method yet, and it's entirely possible that no "best" method (one that is feasible, affordable, and fair [and likely more criteria as well]) exists. In the mean time, we can keep attempting to improve the system even if the next improvement doesn't solve for all of the current problems.
There are, in fact, bendy cucumbers and usually they aren't sold in the markets because they make packaging difficult and are less desirable by the consumers, but it is not like they are forbidden, some markets specifically sell these so less cucumbers are "wasted".
"It's such a fine line between stupid and clever" -- David St. Hubbins, Spinal Tap
Actually, if the guys offered them rohypnol after explaining all the effects and took some themselves it wouldn't be rape. Hell, if they offered them rohypnol after the exact effects of the drug had been drummed into their heads from the time they were in middle school and the guy didn't take any it wouldn't be rape. When intoxicated voluntarily8, unless the person is non-responsive or unconscious it's not automatically rape.
Bzzt, wrong.
https://www.extremetech.com/ex...
On August 18, VW finally admitted to CARB that it had been lying about the cause of the emissions issues for nearly two years. Over the next few weeks, more than 40 employees at VW and Audi destroyed thousands of documents, limiting the ability of investigators to understand how decisions were made and who was responsible for them. As of this writing, VW has paid out more than $22 billion in fines and legal settlements related to its lies, vastly more than it would’ve cost to equip its vehicles with adequate air quality control systems in the first place. VW recently began selling diesel cars in the US again, and such sales represented 12 percent of its total vehicle sales in April, down from a high of ~25 percent in the years before the scandal. It is not clear if demand for diesels in the US will recover. Clearly the cars are still enticing to some buyers, but VW’s brand took a substantial beating throughout this process.
It should be noted that the feds only fined VW about 1.8 billion. Most of that fine came from individual lawsuits and state lawsuits, especially Califonia. You do not fuck with CARB.
Maybe you should learn to read a complete sentence or two, rather than having a meltdown as soon as you see the word "capitalism"?
Il n'y a pas de Planet B.
No wonder you're such an ass. You live in a gated community.
You don't read my comments, do you? I mean, that's cool. I wouldn't want someone who spends time following what I say just to anonymously author cowardly whinges about it to actually know too much about me.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Does this therefore mean "you cannot have good measurable targets"?
I see what you did there. I still can't believe you did it, but I see it.
If only the EU was more like France, getting 75% power from nuclear. Oh wait...