24 hours, no more no less? I doubt it.

March 20, 2009 1:24 PM

Examples of suspiciously round numbers that have, in all probability, been fudged up or down? Extra points for cases where those numbers can be safely discarded in favour of a different, more realistic figure.

We just had a refrigerator delivered and we are advised to let it "stand" for 24 hours before plugging it in. The interwebs confirm this as a widely invoked rule of thumb, the rationale being that in a fridge transported horizontally, oil from the compressor seeps into places it doesn't belong such that the fridge mustn't be switched on until the oil has time to settle back into the compressor. If plugged in too soon, the oil tends to block refrigerant lines (negatively affecting cooling performance) and could potentially cause the compressor to fail completely.

I accept this explanation, and am willing to adhere to the 24 hour guideline just to be safe, but I find it incredibly unlikely that a full day (not one hour more or less) is really required for this process; instead, I would suppose the "true" time it takes would be 12-18 hours at most -- potentially far less -- with the remainder added as a safety margin and to account for idiosyncrasies across brands and models. Instead of providing an accurate figure with each model, a nice, round, extra-safe number that's easy to remember is promulgated by the entire industry.

Another example that springs to mind is the maximum recommended dosage information for non-prescription drugs (e.g. not to exceed 1000mg every 4 hours), which must be set well below the "true" toxic dose for safety reasons and to compensate for the low granularity in the "adults" and "children" doses as opposed a more accurate dosage based on body mass (e.g. 300mg per kg per hour). This fudged number -- say, 40% of toxic for a body mass 1 standard deviation below the mean for an adult or child -- would probably be further rounded down to a multiple of the quantity of drug in each tablet. Alternatively, the tablet size would be adjusted to be a factor of the various fudged dosage guideline(s).

So my question to you: which fudged numbers have you come across? Why do you think they are fudged? What factors would need to be taken into account to determine the corresponding "true" number? Anecdotes where you have personally shown the fudged number to be so? Thanks!

posted by onshi to grab bag (29 answers total)

$700,000,000,000

Quote:

In fact, some of the most basic details, including the $700 billion figure Treasury would use to buy up bad debt, are fuzzy.

"It's not based on any particular data point," a Treasury spokeswoman told Forbes.com Tuesday. "We just wanted to choose a really large number."
posted by Salvor Hardin at 1:30 PM on March 20, 2009 [1 favorite]

The 3,000-mile oil change myth.
posted by MrMoonPie at 1:35 PM on March 20, 2009

@MrMoonPie: That's a good one! Not only are there clear benefits to breaking the "rule" (less frequent oil changes = lower total cost of automobile ownership), it is also clear that following it slavishly has consequences (wasted oil = necessary toxic waste).
posted by onshi at 1:48 PM on March 20, 2009

Err, that's "unnecessary toxic waste".
posted by onshi at 1:48 PM on March 20, 2009

93 million miles to the sun. According to the web it is 92,955,817.7 mi
posted by notned at 1:54 PM on March 20, 2009

@notned: I should have clarified that I'm not so much looking for straightforward rounding (92.955817.7 to 93, a fudge factor of only .0475%) so much as substantially divergent figures.
posted by onshi at 1:57 PM on March 20, 2009

Elevator load limits are always big round numbers. 2,000lbs, whathaveyou. And bridge loads are always in mutliples of half-a-ton (no vehicles exceeding 3.5 tons) and so on.
posted by rafter at 1:58 PM on March 20, 2009

This is not an example, but one of the most fascinating pieces of math I have ever seen is Benford's law, which in essence says for most real sources of data, the first digit is '1' about 33% of the time, and falls off to '9' being about 5%. It's used to discover discrepancies in data all the time.
posted by Lemurrhea at 2:11 PM on March 20, 2009 [3 favorites]

My favorite is the story of Mt. Everest's height (from wikipedia):

In 1852, stationed at the survey's headquarters in Dehradun, Radhanath Sikdar, an Indian mathematician and surveyor from Bengal, was the first to identify Everest as the world's highest peak, using trigonometric calculations based on Nicolson's measurements.[7] An official announcement that Peak XV was the highest was delayed for several years as the calculations were repeatedly verified. Waugh began work on Nicolson's data in 1854, and along with his staff spent almost two years working on the calculations, having to deal with the problems of light refraction, barometric pressure, and temperature over the vast distances of the observations. Finally, in March 1856 he announced his findings in a letter to his deputy in Calcutta. Kangchenjunga was declared to be 28,156 ft (8,582 m), while Peak XV was given the height of 29,002 ft (8,840 m). Waugh concluded that Peak XV was "most probably the highest in the world".[6] In actuality, Peak XV was found to be exactly 29,000 feet (8,839 m) high, but was publicly declared to be 29,002 ft (8,840 m). The arbitrary addition of 2 feet (60 cm) was to avoid the impression that an exact height of 29,000 feet was nothing more than a rounded estimate.[8]
posted by rhizome at 2:21 PM on March 20, 2009 [15 favorites]

RDAs on a nutrition label must be fudged to produce a one size fits all, or rather, the printed RDA is probably only for a specific age-gender group.
posted by Gyan at 2:26 PM on March 20, 2009

Fudged numbers permeate all aspects of life. Fudging is a way of dealing with uncertainty.

When I was a teenager and worked at McDonald's, prepared sandwiches could be kept for 5 minutes, cooked meat for 20 minutes, raw meat in the small freezer 2 hours, raw meat in the walk-in freezer 3 days. The actual number of minutes before the food becomes unsafe to eat is undoubtedly much higher than the cutoff, but you don't actually know what it is, because what if the freezer temperature is wrong, or what if the raw meat sits on the counter for a few minutes before going in the freezer? Because you can't be certain procedures will be followed correctly, the cutoff is well below the danger zone.

Now that I've been to engineering school I would call this a "margin" or a "safety factor". If your bridge can hold 500 cars, you wouldn't design it to hold the weight of 501, because what if one day it's all trucks loaded with anvils, and there's a parade alongside and everybody jumps at the same time? Moreover, what if you design for 600 cars, but someone screws up the welding and you actually fail at 300? Even if the stars align and the worst case occurs, your bridge has to stay up. So you design it to hold 1500 cars even though only 500 will fit.

The whole point behind your 24 hour fridge rule of thumb is that there IS no accurate figure. Nobody can know the condition of the fridge as it arrives at your door with respect to the amount that the oil has moved. Moreover nobody has the resources to do the testing to provide a more accurate figure -- what a nightmare, imagine, every time you design a fridge you need to test for compressor failures after 2 hours, 4 hours, 6 hours etc -- and even if they could, why should they? They only expose themselves to liability by encouraging the customer to operate closer to the razor's edge.
posted by PercussivePaul at 2:38 PM on March 20, 2009

The NY Mets are notorious for faking their attendance figures to make the team appear to be more popular / competitive with the Yankees.

From the NY Daily News' sportswriter's blog:

But now, the Mets have been adding pretend fans in tens of thousands, each and every night. At their last two games, announced attendances of 51,489 and 47,093 weren’t just exaggerations. They were dreamy fabrications....I was at these two games and personally witnessed entire, vast regions of the upper deck devoid of life. This is a stadium that seats 55,601. There were no more than 35,000 fans in the park on Monday, and no more than 25,000 on Tuesday.

posted by jenkinsEar at 2:46 PM on March 20, 2009

I don't see why you use the term "fudge" -as in faked or falsified.

All those numbers are estimated higher/lower than actual because these are generalized numbers and results/effects may vary by product, environment, etc., and it's better to be safe than sorry.

You even stated yourself "willing to adhere to the 24 hour guideline just to be safe".
posted by wongcorgi at 2:59 PM on March 20, 2009

When I used to work at a cable modem ISP, the fix for some problems was to unplug the modem and plug it back in. I worked in the IT department, and I would usually tell people to wait about 5 seconds, just so they wouldn't do a quick out-in that could leave the modem in an inconsistent state.

When my boss passed this number along to the call center manager, he added his own margin of safety, and told her 10 seconds. When she passed it along to the supervisors, she said 30. They told the agents one minute, and the agents ended up telling the customers anything from a minute to ten minutes. Each person that heard and passed along the number thought they were getting the absolute minimum and figured they had better add a little for safety. It wasn't just us, either; I once called my current Internet provider and was told that I should unplug my modem for a full hour to resolve the problem I was having.

One more - if a computer is locked up hard, it won't shut down when you press the power button. The solution is to hold the button for 4 seconds, forcing it to turn off anyway. You can't tell someone 4 seconds, though, because they'll hold it down, count to 4, then let it go and tell you it didn't work, and next thing you know you're arguing over how long 4 seconds is. If you tell them 10 seconds, they'll push it down for at least as long as it takes, then stop on their own when the PC is off.
posted by pocams at 3:20 PM on March 20, 2009 [1 favorite]

So pocams, I have YOU to blame for those countless wasted minutes of my life? I've power-cycled my cable modem more times than I can count or remember, sitting there all scrunched up under my desk, waiting 30 seconds before plugging it back in. (Following instructions from all the customer service agents, websites, etc.) And to think, all of that may go back to ONE GUY. (After a few years I wised up to the fact that all I really had to do was make sure it powered down all the way before powering it back up, and I no longer count to 30. Also, my current cable modem seems to work better than previous ones, so there isn't as much power cycling.)

But, so as to not get deleted, I always wondered about car manufacturers' instructions to drive under 55mph for the first 100 miles, or 500 miles, or whatever it is, of owning a new car. Under 55, precisely? For 100 miles, exactly?
posted by iguanapolitico at 3:30 PM on March 20, 2009

Hard Drives and Computer Memory aren't necessarily rounded off, but the numbers are often a source of confusion:

From wikipedis

"Since the early 2000s most consumer hard drive capacities are grouped in certain size classes measured in gigabytes. The exact capacity of a given drive is usually some number above or below the class designation. Although most manufacturers of hard disk drives and flash-memory disk devices define 1 gigabyte as 1,000,000,000 bytes, the computer operating systems used by most users usually calculate size in gigabytes by dividing the total capacity in bytes (whether it is disk capacity, file size, or system RAM) by 1,073,741,824. This distinction is a cause of confusion, as a hard disk with a manufacturer-rated capacity of 400 gigabytes may be reported by the operating system as only 372 GB large, depending on the type of report."
posted by clearly at 3:30 PM on March 20, 2009

Statistics on penis size. I'm totally serious.

In the 50's, Kinsey did a great deal of work on a comprehensive survey of human sexuality. As part of his research, he asked male subjects to measure the length of their own penises to the nearest quarter of an inch. One blatantly obvious thing that leaps out from the data is that a histogram of self-reported length is mostly bell-curve shaped, but it has major deviations below the gaussian on odd quarter-inch lengths (ie, 5.75, 6.25, 6.75) and deviations above on half-inch lengths, suggesting that participants were rounding to the nearest half-inch.

Also, most subsequent studies to the Kinsey one in which medical professionals did the measuring have reported a smaller average length-- sometimes by a little, sometimes by more.
posted by Maxwell_Smart at 3:43 PM on March 20, 2009

This is kind of an exception that proves the rule, but I've read that Radovan Karadzic (the former rule of the Bosnian Serb state, now on trial for war crimes) used to exploit this phenomenon in order to lie more effectively. Rather than say (lie) "500 people were killed", he'd say (lie) "517 people were killed", on the assumption that the apparent precision would give a sense of veracity to his words.
posted by WPW at 3:52 PM on March 20, 2009 [1 favorite]

The universe is 14 billion years old.
posted by BitterOldPunk at 4:00 PM on March 20, 2009

iguanapolitico, IWAAE (I was an automotive engineer), and the actual requirement is for the engine rpm to vary below a threshold for the first 5-10 hours of operation.

A brand new engine has components of varying tolerances fitting together, and running the car above 55mph could cause the bearings to heat up and seize. Keeping the speed low will make sure that any rough running bearings (and the piston rings and cylinders) to smooth out. This period is known as the break-in period.

(And the 5-10 hour requirement is based on the size of the engine, whether it's diesel or petrol, whether it's supercharged, and whether it's operating a drivetrain. For engines running a fixed load like a generator or construction equipment, the break-in times are larger, heading up to about 50 hours).
posted by Arthur Dent at 4:01 PM on March 20, 2009

@ PercussivePaul: I disagree with your assertion that "there IS no accurate figure" for the amount of time it would take for the oil to drain back into the compressor where it belongs. The manufacturer of my fridge could very easily test this by mouting the relevant components on a pivot, rotating them to horizontal (or past horizontal if this would produce an even "worse" initial state) until all of the oil had moved as far into the refrigeration lines as possible, then rotating back to vertical and observing how long the oil took to drain back to the compressor. Given that residual oil is widely known to cause complete failure of the appliances they manufacture, it is not unreasonable to expect that they would have an interest in determining this with some accuracy rather than simply going with 24 hours, especially when this causes considerable inconvenience to customers (i.e. I am losing food to spoilage during this 24 hour period that might be salvageable after just 6 hours in a cooler). Either that or they should design a compressor that's not vulnerable to this issue, though that sound so obvious that it must be prohibitively difficulty or expensive. They may not have bothered to test for an accurate figure, but that hardly means it would be difficult to obtain one.

@ wongcorgi: The reason I am adhering to the 24 hour guideline "just to be safe" is because I do not own the fridge. It belongs to my landlord, so I'm not going to experiment with it. If I had purchased it myself, I would consider plugging it in sooner. I agree that using 24 hours as a general rule across all refrigerators isn't so much fudging as a useful guideline, but when it's given in the instructions provided with a specific appliance as though it were accurate for that model then I do consider it to have been falsified. They set the number not by taking a "real" estimate for the appropriate time and then adding a reasonable margin but rather by picking a whole day.
posted by onshi at 4:50 PM on March 20, 2009

Well fine, let's say it is possible to precisely determine that fridges take 6 hours to drain, not 24. The fridge company then says, OK everyone, it's 6 hours! Users will then drain the fridges for 4 hours, plug them in, and BOOM. It's not just uncertainty on the failure mode this figure deals with, but uncertainty as to whether the procedure is actually going to be followed.

But that was just an example, no need to get hung up on the fridge. As far as I can tell, you want to know why number-fudging like this happens. I think my explanation -- reducing the risk of failure in the face of uncertainty -- is pretty reasonable. What else are you looking for by asking this question?
posted by PercussivePaul at 5:34 PM on March 20, 2009

I'd be fine with 6 hours plus margin equals guideline. What gets my BS-detector going is the suspiciously roundness of 24 hours. I am hoping not for a general explanation of why margins might be added (as I think I showed in the question, especially with pharmaceuticals, that I understand this necessity) but rather for interesting examples of suspicious, probably-fudged numbers passed off as relatively precise.

The Mt. Everest one was pretty neat, even though it was kind of the opposite (round number made otherwise to enhance credibility).
posted by onshi at 5:47 PM on March 20, 2009

More than likely, your refrigerator manufacturer has a very good idea of what the top end limit is for oil to resettle. They probably have done very similar tests to the one you describe. But the behaviour of your fridge is not the only thing that matters.

By setting it to a day, you'll guarantee that the vast majority of your customers will, at minimum, leave it overnight. Set it to any time less than that, and they're going to start thinking about how they can fudge the time down so that they can get that puppy plugged back in tonight before they go to bed.
posted by jacquilynne at 5:47 PM on March 20, 2009 [1 favorite]

But then this is chatfilter. We could list examples all day long.

By the way I disagree that "1 day" is suspicious and passed off as precise. I think it is quite clearly an approximation. If it said "7.5 hours" I would think it's being passed off as precise.
posted by PercussivePaul at 5:57 PM on March 20, 2009

@PercussivePaul, it might be obviously an approximation to you, but, trust me, it is not to 98.23% of the population. (play on numbers notwithstanding)
posted by Precision at 7:32 PM on March 20, 2009

I'm going to quibble on the "3000 mile oil change myth" myth. The internet and barrooms are full of geniuses who have decided that they are getting away with something by extending their oil changes out to x months or 10,000 miles or something like that.

The reality is much more complicated. First, certain oils thicken as they age. You suffer worse and worse fuel mileage as the oil ages. Second, oil is meant to pick up and contain dirt in solution so that it isn't sludging up the inside of the engine. Third, if you say 3000, you will probably end up with most people doing it every 6000 because we are lazy. Fourth, oil is cheap insurance. I'd rather have a car that I tearfully sell because its body has turned to dust after 15 years than a car that I have to scrap because I cheaped out on oil changes and it has spun a bearing. Or drives like a go kart because the engine is worn. Fifth, oil from oil changes gets recycled, so we really aren't wasting anything. If you count the gas savings from not driving around with molasses in the crankcase, it's a net win. (Think about it- a car's crankcase holds what, a gallon of oil? 3000 miles is 100-200 gallons of gas, produced from lord knows how much base stock.) Sixth, if you read the owners manual, it usually says something like 7500 miles for "normal" service, 3000-5000 miles for "severe" service. If you look at their definition of "severe", it pretty much comprises anything besides cruising on the highway at 55 mph. Seventh, oil gets contaminated with all manner of gunk from the air and the gas it burns, forming all manner of compounds that will eat your engine.

Sure, maybe 3000 is a made up round number. I knew a guy whose car needed the oil changed every 1000 miles or it wouldn't idle right. And I'm sure there are Buicks out there that go 10000 miles before the oil life monitor dings. But 3000 is a nice round number that's easy to remember that for the most part guarantees that the engine will be well maintained in situations where owners don't have some other system for monitoring oil life.

/rant off
posted by gjc at 9:08 PM on March 20, 2009

In my job as the HR guy...I came across a LARGE group of people who clocked in and out of work at the exact right times. 8:00, 12:15, 3:45....

I noticed this for a few months before I asked IT to intervene and get me more info. How were these people clocking in and out at the perfect times?

I found out that the controller/accountant of the company was shaving hours. BIG TIME!

End result...the company got sued for a quarter of a million dollars.

If only the accountant would have figured that having people clock out at 5:02 would be better than 5:00 when changing it from the original 6:37. If only...
posted by hal_c_on at 2:24 AM on March 21, 2009

erm ... the 3000mile oil change is certainly a good idea for a number of older automobiles ... and history is probably where it comes from.

New automobiles ... no question that 3k is on the short side.

VW Bugs, without oil filtration, where the oil also provides cooling to the engine ... 3k was about the max ... and the factory recommendation was lower in dusty environments (sometimes less than half that). Ignore the factory recommendation significantly and your bug would have increased engine problems, lower lifespan, and big repair bills (earlier han otherwise)
posted by jannw at 7:42 AM on March 21, 2009