October 14, 2004

11:11 AM

Part One: I've heard people say that the "close doors" buttons in elevators don't work. They certainly don't seem to do anything in my building. Do they not work because of poor design or are they literally (purposefully) not hooked up to anything?

Part Two: Are there other examples of mechanical (or software) placebos? I've heard that the "progress bar" you see when Windows boots doesn't actually mean anything. Microsoft just discovered that people were happier if there was some sort of animation going on while their computer was starting up.
posted by grumblebee to (44 comments total)

Sometimes they do nothing at all, sometimes they are on a minimum-cycle timer, which isn't much shorter than the standard close time.

*Except in fire mode*.

That's what they're *really* there for: in fire mode, the doors do not open or close automatically -- it wouldn't be good for a fireman to go to a floor, and have the door open without having time to check if that's going to burn him down where he stands.

Or something like that. ;-)
posted by baylink at 11:20 AM on October 14, 2004

I've heard that many office thermostats are placebos. The one in my former office certainly never seemed to do anything.
posted by Turd Ferguson at 11:20 AM on October 14, 2004

Part One: I've seen both functional and non-functional buttons. The ones in my office building definitely do nothing. The one in my old apartment building definitely worked.

Part Two: Huh, that's interesting. The Mac startup bar always puzzles me because it itemizes what's going on above it (Starting Apache/Starting Network Filesystem/etc.), but the progress bar itself doesn't move in sync with time, nor does it seem to allocate a fixed length to each "task" it runs through. The upshot is that it often spends most of it's time at 90%, which totally bugs me.
posted by mkultra at 11:23 AM on October 14, 2004

It was determined recently that 2 out of 3 of the buttons at crosswalks in NYC aren't hooked up to anything. They used to work, but advances in the system antiquated them (if that's a verb), and now they're placebos.
posted by o2b at 11:24 AM on October 14, 2004

While elevators do differ, the instructions in the elevator at work say that the "close doors" and "open doors" button work when a firefighter has used his key to change it to an emergency mode. At an old job, the "open doors" button worked on the relatively ancient elevator to hold the doors open. "Close doors" didn't do anything.

About progress bars: I always found the Windows 2000 startup progress bar that started half full to be quite funny. Also, while contacting a server, the progress bar in the bottom of Internet Explorer in Windows doesn't mean anything. It just goes up to make you think it's doing something other than waiting for the server.
posted by zsazsa at 11:25 AM on October 14, 2004

I find that the close doors buttons in elevators tend to work when the doors have been held open. I.e., pressing the open doors buttons buys you a few seconds, but immediately pressing the close door button will cancel that out. They don't seem to generally have any athority over the 10 seconds or so the elevator remains open after it reaches a floor.

As for other mechanical placebos, the fuel gauges in most cars lie a good bit.
posted by tirade at 11:53 AM on October 14, 2004

The close-door buttons in some of the hospital elevators where I work are definitely useful - they eliminate the pause with doors open on stopping at a floor. The optical sensor overrides this for safety purposes, of course.

I've also seen elevators with the panel off and asked the repair guy; once I was shown a 'placebo' close door button, just a normal momentary switch with nothing attached to its terminals.

Of course, the same fat-assed people who press both up and down buttons to summon the elevator, then ignore elevators going in the direction they're not going, tend to park said lard ass right in front of the elevator panel, making it impossible for me to do my trick of hovering over the close-door button. I work on the 19th floor and the novelty of 10 minute elevator rides 4 times a day has worn off.
posted by ikkyu2 at 12:02 PM on October 14, 2004

The best progress bar was the one on Apple's GS/OS. It was an interrupt-driven progress bar that moved smoothly through the amount of time it took your computer to start up the last time you booted it. Thus, the only time it wasn't right on was when you made some major change in your system, such as putting a SCSI accelerator card in that made booting twice as fast. (But it'd be back to being accurate the boot after that, of course.)

The elevators where I work have functioning Close Door buttons, but they don't always work depending on circumstances. For example, if the elevator is at the lobby, it will always leave the doors open for several seconds after you hit a floor, just in case someone else is coming in, up to some time limit after which the doors start wanting to close immediately after you hit a floor. During this time period Close Door is a no-op. (If there is more than one elevator waiting in the lobby, however, the doors will start trying to close immediately after you hit a floor, and Close Door works fine.)
posted by kindall at 12:04 PM on October 14, 2004

I believe it was Douglass Copland who suggested that the common cold is transmitted through contact with the "close door" buttons, which is why stressed-out people tend to get sick more often.

I like that theory.

Anyway, I work at a hospital with many buildings and many different types of elevators of all ages. In some elevators the buttons close the doors right away, in others they do nothing except maybe when the firefighter key is instered.
posted by bondcliff at 12:19 PM on October 14, 2004

The illumination of the 'check engine' light on my car's dashboard display doesn't have any relationship with anything going on in the engine, according to a sampling of mechanics who've checked the engine at various times when it's come on. Not sure if it's a plecebo, or just a tease.
posted by normy at 12:32 PM on October 14, 2004

Here's another: The very shallow depth of tread on bicycle tyres for road use. Consumers think tyres should have tread, so the manufacturers give them tyres with a tread pattern on the surface. Consumers generally won't buy completely smooth, or treadless, bicycle tyres, even though the tread pattern serves no purpose and tyres would be cheaper to make (and therefore buy) without it.
posted by normy at 12:40 PM on October 14, 2004

The door buttons on the tube in London are there as a placebo. They are actually activated by the driver. That makes for a bit of fun when the driver has dozed off and everyone is hitting the door buttons because they are stuck there :D
posted by floanna at 12:42 PM on October 14, 2004

If you're like me, you like to squeeze every last mile you can out of your tank of fuel. If you could get 20 miles extra from each tank, that could save you two or three trips to the gas station over the course of a year.

How do you save gas by letting the tank run to the bottom...? It's not like that fuel disappears if you refill the tank- or am I missing something?
posted by drezdn at 12:53 PM on October 14, 2004

Are any software progress bars of any use at all? I always assumed they were all a big placebo conspiracy. (Great question, btw.)

Think about it. What exactly are they telling you? They certainly don't indicate time left...if they did, they could only be accurate if the task "knew in advance" how long it was going to take, the way a second hand goes as fast as it does because it "knows" it the trip around the clock will take one minute. That's impossible. Moreover, if they indicate time remaining, why don't they move at a steady pace, instead of parking at halfway for four minutes, and then suddenly zooming to 98% complete, and then finishing.

If they are telling you what % of the task is complete, measured in some way other than time, such as "files installed," or "files unpacked," then who cares? What good does it do to tell me that 50% of the tasks are complete, if they are the fast 50%, and the second 50% will take three times as long?

I've never seen anything written about the theory behind progress bars, and I'd like to know more.
posted by stupidsexyFlanders at 12:57 PM on October 14, 2004

normy: Not buying it. Treads increase traction on wet roads, and prevent what would in the worst case turn into aquaplaning (though that's unlikely to occur to any serious degree with normal bike use).

drezdn: You save petrol by not lugging around as much petrol.
posted by fvw at 1:01 PM on October 14, 2004

You're not saving gas, you're saving the time it takes to do two/three extra fillups a year, time which will immediately be lost the first time you run out of fuel trying to squeeze an extra mile or two out of it...
posted by Keith Talent at 1:01 PM on October 14, 2004

The writer mentions saving trips, not buying less fuel.

But, also, running an emptier tank will marginally increase efficiency (less weight), and fewer fuel-ups means fewer starts, which also will negligably reduce the amount of fuel used.
posted by Kwantsar at 1:02 PM on October 14, 2004

stupidsexyFlanders: While they're certainly not ideal (and some are very badly designed), even if measuring in time isn't possible, partial information is better than no information at all. And it's easier to tell whether the application or system in question has crashed or hung.
posted by fvw at 1:07 PM on October 14, 2004

Are there other examples of mechanical (or software) placebos?

Well, to bring this to a human level, the Family Courts of the UK and Japan are placebo bodies, to give the public the impression there is independent arbitration to fall back on in the event of family problems.

this is a kind of "no truer word said than in jest" kind of thing...feel free to laugh before you criticise!
posted by SpaceCadet at 1:23 PM on October 14, 2004

Not buying it. Treads increase traction on wet roads, and prevent what would in the worst case turn into aquaplaning

See what I mean?

What you say is absolutely correct for larger, heavier vehicles that use larger tyres at lower pressures - cars, for instance. But bicycle tyres are not car tyres. Car tyres have tread because, in wet conditions, water has to be given a way to flow from under the tyre's contact patch, allowing rubber to remain in contact with pavement. This is because the contact patch of a car tyre is much larger than that of a bicycle tyre - the water has much further to travel. Looked at another way, physics dictates that the pressure forcing water from below the car's tyre is equal to the pressure of inflation - typically 30-40 psi. With a bicycle road tyre, inflation pressure is typically 80-120 psi.

So, comparing a car to a bicycle, with a bicycle, we have a much smaller contact patch and much higher pressure. These are the physical parameters of concern in aquaplaning. It's possible to calculate the speed required to aquaplane a bicycle tyre, and figures in the region of 90-100 mph have been the result.
posted by normy at 1:30 PM on October 14, 2004

You save petrol by not lugging around as much petrol.

I think I read somewhere that running around with a near-empty gas tank is bad for the car, because rust and residue that collects at the bottom of the tank ends up getting into the engine, making a mess of fuel lines, injectors, etc.
posted by swift at 2:08 PM on October 14, 2004

as well, I was told not to run at near-empty, as the gas acts as a coolant for the fuel pump, so running at almost-empty can cause the fuel pump to overheat.

I have a 15 year old car, though, so ymmv.
posted by sauril at 2:16 PM on October 14, 2004

The illumination of the 'check engine' light on my car's dashboard display doesn't have any relationship with anything going on in the engine, according to a sampling of mechanics who've checked the engine at various times when it's come on. Not sure if it's a plecebo, or just a tease.

Not a placebo or tease, but part of the Federally mandated On-Board Diagnostics II (OBDII) system. Check Engine (aka the Malfunctional Indicator Light, or MIL) illuminates when a problem is detected and the corresponding diagnostic code is set.

However ... some codes are erased after X number of engine starts if the problem is not detected again. In plain English, you can wait a few days to see if a constantly illuminated Check Engine light goes out by itself.

On the other hand, if Check Engine ever flashes on and off, you have a potentially serious problem and should stop driving the vehicle to have it examined ASAP.

(Sorry for the blink tag ... I couldn't resist the most appropriate opportunity I may ever have to use it!)
posted by pmurray63 at 2:24 PM on October 14, 2004

Way too much information about progress bars, because I happen to have been working on this recently:

For very predictable tasks, progress bars can be accurate for time remaining. Say you're copying a large number of files from one disk to another. The rate at which the copying occurs is (almost) perfectly steady and unlikely to be interrupted, so as long as the programmer took the time to have the software calculate ahead of time the total size of whatever is being copied, the progress bar can be (almost) perfectly accurate.

For less predictable but consistent tasks, they can be accurate in terms of the percentage of the job complete but not in terms of time. Example here would be downloading a large file: the task itself is predictable and constant, so you can see how much of the job is finished, but the vagaries of network lag means that any prediction about the time remaining is likely to be wrong. It's still possible to make some predictions for remaining time based on how long the first N percent of the task took, so the prediction will grow more accurate as the job progresses -- but this takes extra effort for both the programmer and for the software, so is unlikely to be implemented except for tasks which take a really long time.

For unpredictable tasks which contain lots of different processes, like booting a computer, the only way to be accurate is to cheat. You can't really predict how much time each part of the job will take without actually doing it, and you can't make any predictions based on part X about how much time part Y will take. So the programmer has three options:

• Just up the progress bar an arbitrary amount each time a particular task is completed -- this is what causes the jerky, unpredictable progress bar in windows bootup. This is a real progress bar, in that it will hit the end exactly when the job does, but how far along the bar has moved doesn't really mean much.


• Fake it: guess how long you think the task will take, make the progress bar move smoothly at that rate, and remember to cap it at 100% in case the job takes longer than you expect. (Nothing screams quality like a progress bar that keeps going right off the edge of the screen.) The progress bar in this case isn't actually tied to the job at all, it's just a stopwatch which you hope will finish more or less when the job does. Kindall describes the best-case situation for this, where the "guess" can be based on how long the task took last time you did it; that's a nifty technique and I'm stealing it now -- but it's still not really measuring actual progress, it's just faking it.


• Cut bait, and use a non-progress progress bar, like the OSX barberpole, which just spins to tell you it's still working, with no feedback about how far along it is.

posted by ook at 2:27 PM on October 14, 2004

I have a 15 year old car, though, so ymmv.

Literally...
posted by esch at 2:28 PM on October 14, 2004

great, ook, thanks!
posted by stupidsexyFlanders at 2:32 PM on October 14, 2004

Are there other examples of mechanical (or software) placebos?

I did read of a building whose occupants were complaining about how long the lifts^H^H^H elevators took to come to their floor, until mirrors were installed beside each lift. Now they had something to occupy their time, and they agreed the lifts came quicker (which was bollocks).
posted by John Shaft at 2:45 PM on October 14, 2004

This usually opens a HUGE can of worms, but...

Images for the web do NOT need to be made at a resolution of 72dpi. They are just as good (and just as bad at 300dpi, 600dpi, 5dpi or a zillionDPI). DPI stands for dots per inch, and the idea of an "inch" is meaningless on a computer screen (an image that takes up one inch on my screen may take up three inches on yours). DPI is a command to the PRINTER, telling how many dots of INK to plop down per inch of paper. DPI is meaningful for print, but not for web.

Still, many people think that web images must be 72dpi. Because SO many people think this, apps like Photoshop and Fireworks default to 72dpi. The "placebo" keeps people happy.

(If anyone knows where the 72dpi myth came from, I'd love to hear about it. I've been told that the original Macs that were in many design shops in the early 90s had 72pixels-per-inch on their screens. But I don't know if this is true.)

If you don't believe me, try making two images in photoshop: make one 100pixels by 100pixels at 72dpi and the other 100pixels by 100pixels at 300dpi (the only difference being the dpi). Check the file size of both: the same. View both on a web page: they both look the same (same size, same quality). Now print both: very different!
posted by grumblebee at 2:45 PM on October 14, 2004

I had a Nissan Stanza that had a dashboard light that said "SENSOR". It came on every 30,000 miles to remind you to check the exhaust sensor. It stayed on until you tore off half the plastic covers under the dashboard and found the really tiny, inconveniently located button to make it go off.

After 90,000 miles, I fixed the problem with a piece of black gaffer's tape cut to size.

The door buttons on the tube in London are there as a placebo. They are actually activated by the driver.

Doesn't the driver basically just hit a button that makes those buttons work? Because if you're in a sparsely-populated car and passengers aren't using every door, the others won't open.
posted by Vidiot at 3:20 PM on October 14, 2004

If anyone knows where the 72dpi myth came from, I'd love to hear about it. I've been told that the original Macs that were in many design shops in the early 90s had 72pixels-per-inch on their screens. But I don't know if this is true.

It is.

The basic idea was for one pixel to roughly equal one typographical point, so that what you saw on the screen would be very close to what you got on the printout.
posted by kindall at 3:35 PM on October 14, 2004

grumblebee, Macs have always run at a nominal display resolution of 72 dpi, clear back to 1984. Back in the days of monochrome nine-inch monitors, there actually were 72 pixels per inch, chosen because there are 72 "points" (unit of typography) in one inch; you could print a document on your ImageWriter, hold it up against the screen, and the outlines would (basically) match up. More usefully, you could pick "12 point" text from the menu, the software would draw it 12 pixels tall, and everyone would be happy. This was revolutionary at the time, and when the whole "desktop publishing" thing kicked off all those apps assumed that the display was set to 72 pixels per inch. It didn't stay true for very long (the Macintosh II let you set your monitor to an arbitrary number of different resolutions), but the notion that 72 dpi is "screen resolution" stuck around.
posted by Mars Saxman at 3:48 PM on October 14, 2004

A lot of weapon systems will show some sort of progress displays to the pilots/operators simply to give them a warm fuzzy. The display isn't necessarily tied to anything that they can control or even anything informative -- it just assures them that something is happening and the system hasn't crashed.
posted by joaquim at 3:49 PM on October 14, 2004

The "close door" buttons in Malaysia and Japan seemed to work just fine. And this was during "normal" mode of operation -- that is, the initial opening of the doors to accomodate all of the invisible people thus the pushing of the button, and the astounding that it actually worked.
posted by user92371 at 4:19 PM on October 14, 2004

My parents have an Oldsmobile on which the brake pedal seems to be purely cosmetic.
posted by adamrice at 5:17 PM on October 14, 2004

They work in my office building,
and as for this: It was determined recently that 2 out of 3 of the buttons at crosswalks in NYC aren't hooked up to anything. They used to work, but advances in the system antiquated them (if that's a verb), and now they're placebos I can only say, "I knew it!!!"
posted by amberglow at 5:53 PM on October 14, 2004

And it's easier to tell whether the application or system in question has crashed or hung.

precisely. also sometimes can be helpful in determining where (eg reading/writihg which particular file) an application has crashed or hung.
posted by juv3nal at 6:05 PM on October 14, 2004

Though, FWIW, there are actually 72.27 points in an inch.

The elevators/mirrors thing comes from Conceptual Blockbusters, IIRC. Either there or The Mythical Man Month. They're on much the same patch; I can't remember exactly which.
posted by baylink at 6:07 PM on October 14, 2004

I also read an article stating that most crosswalk "Push to Walk" buttons aren't connected to anything anymore, particularly those in the Bay Area. But, it's too expensive too remove them all, people like them for the sense of empowerment, so there they sit.
posted by robbie01 at 8:09 PM on October 14, 2004

I used to work in an office where people defragged their hard disks once a week because the IT staff told them it was a good idea. People loved watching the defragger "tidy" things up, wasting an hour a week to save maybe a fraction of a second on disk accesses.
posted by teg at 9:17 PM on October 14, 2004

normy: Agreed, true, across-the-tire aquaplaning won't occur on a bike. But there will be areas of reduced traction. At least that's the only explanation I can think of that would explain the fact that my old bike's tires when they were worn down got rather dangerous in wet weather on tight turns (luckily someone stole it at that point so I didn't get into any interesting accidents).
posted by fvw at 5:52 AM on October 15, 2004

More likely what happened was the rubber compound hardened with use/age and became less grippy.

Different car tires have different hardness and therefor wear resistance and grip built in. At the extreme ends of the spectrum you can buy tires that will pick up stones when you drive over them but with stick like ugly to a bear and tires that are so hard you can't cut them with a knife but they will last for 160,000 kms and won't be punctured by rocks..
posted by Mitheral at 10:13 AM on October 15, 2004

The door buttons on the tube in London are there as a placebo.

They can actually be turned on so they do work. However, because of concerns about possible lawsuits from people injured by others using them they currently chose not to make them available.
posted by kerplunk at 12:11 PM on October 15, 2004

On the crosswalk note: My little sister did a science project in grade school (this is about 10 years ago, near San Francisco) where she proved that almost none of the crossing buttons did a thing--and, if I remember correctly, she even isolated the variable of pressure pads possibly being in use in the roadway. Afterwards, she interviewed the guy in charge of all the stoplights at the DMV. He buckled when presented with the evidence, and admitted that most of them were never hooked up in the first place.

As an aside, we always called them "pedestrian happy buttons" when I was a kid.
posted by armchairsocialist at 6:36 PM on October 15, 2004

Well, to bring this to a human level, the Family Courts of the UK and Japan are placebo bodies, to give the public the impression there is independent arbitration to fall back on in the event of family problems.
Add Australia to that list.
posted by dg at 8:17 PM on October 15, 2004