Wikipedia:Reference desk/Archives/Science/2015 May 11

Science desk
< May 10	<< Apr | May | Jun >>	May 12 >

Welcome to the Wikipedia Science Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

May 11

Fish identification

A friend of mine has two of these fish in her (non-salt water) fish tank. She was told by the pet store that she got them from that they were a particular type. I think she was told the fish were gold barbs but I'm not sure of that. She was told that they'd be rather docile but they've killed nearly every other fish in the tank and even killed the snails that were once in the tank. They are also larger than she was told they would get. Their length is maybe about 3.5-4 inches. The slightly fuzzy images are here and here. Sorry about the quality. They were moving very quickly. Can anyone tell me what kind of fish they are? Dismas|^(talk) 04:27, 11 May 2015 (UTC)

They could certainly be gold barbs which are highly bred and variable; the fish have the cyprinid (carp-related) body form and the typical barbs on either side of the mouth. These just aren't very brightly colored. That can come from genes, food, and/or environment. The problem is they are notoriously aggressive, per google. μηδείς (talk) 05:34, 11 May 2015 (UTC)

Names of internet trunk lines

I noticed on this map that the internet trunk lines are all named after famous explorers. Are these really the official names? Or just something the author came up on their own? My other car is a cadr (talk) 05:58, 11 May 2015 (UTC)

As far as I can tell they made the names up. Greg's Cable Map, which is their only cited source for the cables, doesn't use those names. They also seem to have made up the lines themselves. Also, apparently there's no Internet surveillance in Canada, and the US imprisons people whereas most other countries don't? Or something? I deleted the image from the one Wikipedia article where it appeared. -- BenRG (talk) 07:42, 11 May 2015 (UTC)

I agree with BenRG. The map does not correspond - not even in abstraction - to any other reliable source I've ever read. Consider reading list of international submarine communications cables. That list names the companies who administer the cables and you can follow up on each line in much greater detail. The cartoonish "subway-map" doesn't even seem to have connections in the right places! Nimur (talk) 10:18, 11 May 2015 (UTC)

IMNSHO that map is a complete waste of bytes. Roger (Dodger67) (talk) 10:28, 11 May 2015 (UTC)

Maybe a high-tech version of the fictitious "Train leaving on Track 5 for Anaheim, Azusa and Cucamonga." ←Baseball Bugs ^{What's up, Doc?} carrots→ 17:15, 11 May 2015 (UTC)

Cats like earwax

Why do cats like licking earwax? Pickuptha'Musket (talk) 11:57, 11 May 2015 (UTC)

Your premise is faulty. Cats lick their butts too and I doubt they like it. It is about Personal_grooming.196.213.35.146 (talk) 12:43, 11 May 2015 (UTC)

I meant human earwax! LOL Pickuptha'Musket (talk) 12:47, 11 May 2015 (UTC)

See cat communication, social grooming, and e.g. here [1]. Cats groom each other, and domestic cats often groom humans. Some cats go for an ear, others might go for a human's nose or hair. You might be interested in these books on human-animal interaction [2] and domestic cat behavior [3]. Google /cat groom human/ for more perspectives on the topic. SemanticMantis (talk) 13:36, 11 May 2015 (UTC)

• See the complex question fallacy. The OP has posed us a question which is literally unanswerable in its current form. It takes the form "When did you stop beating your wife?" The OP's question requires us to assume to be true a statement which has not been shown to be so yet. Namely, that cats enjoy licking earwax. Since we don't know if that statement is true or not, we cannot possibly provide an answer as to why it would be true. First establish that it is true, then we can look for references to answer why. Otherwise, there's no point in answering it. --Jayron32 14:19, 11 May 2015 (UTC)

A quick search reveals a fair amount of online discussion about this behavior, including some videos documenting it, so the premise of the question seems sufficiently well established. It is likely that Wikipedia lacks an article on this and that we will be unable to provide a link to a reliable source explaining it, and so the questioner will have to be satisfied with unreferenced discussions on other web sites. -- ToE 14:51, 11 May 2015 (UTC) OR: I was unable to duplicate the behavior with my cat and my ear wax.

The suggestion is that it smells foody see here. DuncanHill (talk) 15:03, 11 May 2015 (UTC)

The most obvious immediate response to the OP in a case like this would be, "Who says they do?" Some of you have now answered that question. ←Baseball Bugs ^{What's up, Doc?} carrots→ 17:11, 11 May 2015 (UTC)

Or perhaps cats have qualia that we don't have for "groomy". Count Iblis (talk) 18:29, 11 May 2015 (UTC)

• It would seem the OP hasn't yet realized that when his cats attempt to eat his ear-canal contents they are actually hungry for brains. See the article zombie-cat. μηδείς (talk) 00:42, 13 May 2015 (UTC)

If I have twice the metabolic rate of a normal person then does that mean I have twice as much brain power?

My laptop of similar design as another laptop is much more powerful, because it consumes 50 watts when using all its 4 cores at maximum power compared to the 15 watt laptop that I decided not to buy. I was wondering if the same can be said about my brain. Someone of my weight and height has a metabolic rate of about 1500 Kcal per day, however, my metabolic rate is about 3800 Kcal/day. If I take into account the 1000 kcal due to exercise, then I'm still at about twice the metabolic rate of a normal person of my weight and size. Since the brain is supposed to use about 20% of the energy used by the body, does this mean that my brain is capable of working twice as hard as the brains of other people? Count Iblis (talk) 18:27, 11 May 2015 (UTC)

Doubtful. There's no reason to think an increased metabolic rate means every aspect is increased proportionally. You might move more than the average person, or keep the room cooler, requiring burning more energy to stay warm. Also, wouldn't it be a measure of your current energy usage, including your brain, not the maximum possible ? StuRat (talk) 18:32, 11 May 2015 (UTC)

Perhaps your brain is only half as efficient! ( No insult intended.)  :) Dbfirs 19:38, 11 May 2015 (UTC)

That would mean it generates excess heat, making him a hothead. :-) StuRat (talk) 19:51, 11 May 2015 (UTC)

The section Neuroscience and intelligence#Neural Efficiency suggests that more intelligent individuals have a higher neural efficiency, using less energy per task. Also if one measures metabolic rate by food intake, rather than a thermal measurement, it may not be a higher metabolism, just less efficient food digestion. Or a large tapeworm. --Mark viking (talk) 19:54, 11 May 2015 (UTC)

The statement about CPUs is dubious. This CPU with a TDP of 47W is probably less than twice as fast on highly parallel workloads as this CPU with a TDP of 15W, and for serial computation it may actually be slower. (If I wasn't restricted to CPUs released at the same time, I could have picked this one.) -- BenRG (talk) 20:18, 11 May 2015 (UTC)

Yeah it depends on the specific workload but performance/power of CPUs are fairly variable. You could easily find a CPU of similar generation with higher power consumption but lower performance under many workloads if you're comparing AMD to Intel, and even within Intel you could find it if you choose carefully enough. Nil Einne (talk) 11:44, 12 May 2015 (UTC)

Surprised no one has pointed out comments on one's brain performance are questionable if you think it's normal for anyone to have a metabolic rate of 15 Kcal per day. (Do people in a coma even have a rate that low?) Or the same if you think that 3800 Kcal per day - 1000 Kcal per day is only about twice 15 Kcal per day. ;-) Nil Einne (talk) 11:51, 12 May 2015 (UTC)

• Count Iblis is either a demon or about to die of a very high fever, given a doubled reaction rate in many everyday systems implies a 10 degree Celsius increase in (body) temperature. Or maybe he's a warm vampire? A doctor should know. Moriarity, Faust and Jeckyll specialize in the field. μηδείς (talk) 00:48, 13 May 2015 (UTC)

He's a Count - perhaps Dr van Helsing would be more appropriate. {The poster formerly known as 87.81.230.195} 212.95.237.92 (talk) 13:10, 13 May 2015 (UTC)

Yes, exactly. Conservation of energy applies to the human body just like any other system. Energy comes in as calories and goes out as heat and (to a degree) kinetic energy during exercise. The energy from that food can't simply vanish. One possibility is that it's being pooped back out again...but that would definitely imply some kind of major medical problem that we can't go into here. For sure it isn't coming out as heat - because (as Medeis points out), unless you're living in an ice-box and would die if unable to produce that extra heat, the only way to radiate more heat than other people nearby is to have a higher body temperature...a MUCH higher body temperature in fact. So I very much doubt Count Iblis' data. The argument that the energy is going into powering your brain (and the implication that this makes you somehow much more capable than other humans) is bogus because 100% of the energy that goes into your brain goes our again as heat...so again, only a dramatic (and exceedingly unhealthy) increase in body temperature compared to other people could possibly explain these results.

I don't know how you get the idea - but your body can't break the laws of thermodynamics - the excess energy that you consume has to go somewhere. In most people, the energy from excess food intake is stored as fat - but the only way you can be consuming 2x the number of calories of other people is to be hotter (dangerous!) to have a much larger skin area to radiate the heat from (I think you'd know if that were true) - or if you wear light clothing in a frigid environment - or if you're doing a truly crazy amount of exercise to shed the excess energy as motion of some kind.

So let's think about the exercise angle for a moment...let's suppose you're a 200lb body weight - the heavier you are, the more calories you burn...so this is a generous assumption...if you're burning 1800 excess calories - what level of exercise would that demand? Well, let's suppose you're running - at a 10 minute mile pace. That burns 930 calories per hour - so to shed 1800 calories per day, you'd have to be running for 2 hours - or around 12 miles per day. Is that the case? Other options are equally difficult - playing basketball is 600 calories per hour - so three hours of basketball per day? Climbing a hill with a 40lb backpack burns 800 calories per hour...so two hours of that wouldn't be enough. Also, bear in mind - these are EXCESS amounts of exercise - over and above what a normal person would be doing...and if you only weigh 130lb - you'll need to do about twice the amounts I described above.

Some people claim that their gut bacteria are consuming the excess calories - but they too would have to either store the energy or exhibit huge heat output - which would increase your body temperature (or, more likely, reduce the number of calories you need to maintain a normal body temperature)...either way, what goes into your body has to come out or be stored...there are no other alternatives.

Bottom line: This isn't happening as you think it is - the laws of physics don't allow it...maybe you're eating less than you think...or you're putting on weight...or you have some kind of nasty digestive disorder ("No Medical Advice - Go see a doctor")...or you're running a perpetual high fever ("No Medical Advice", etc, etc)...or you're some kind of extreme exercise fanatic...or you're a member of an antarctic nudist colony...or something like that. But for 100% sure it isn't because you live a 'normal' lifestyle with 'normal' body temperature and exercise regime, but have an incredibly impressive brain that's thinking orders of magnitude faster than anyone else - because that would show up as hugely elevated body temperatures - and running that high of a fever for that long would be enough to kill you.

So, examine your assumptions and figure out which of them is wrong...one of them must be.

SteveBaker (talk) 18:03, 13 May 2015 (UTC)

A higher resting metabolic rate doesn't imply a fever because with more muscles you also have a more vascular body that can transport heat more efficiently. Also the lower total fat percentage also means that you have less insulation. My total fat percentage is about 13% according to my weighing scale (not sure how accurate this measurement is, though). Now, just compare that to a more typical value of 20%; that 7% difference would be about 4 kg worth of fat which is more than the weight of my winter gear that I used this winter in the Arctic. Now, the insulation of all that body fat may be less effective than my winter gear, but it is likely to have a significant effect. Count Iblis (talk) 23:21, 13 May 2015 (UTC)

You seem to be describing an athlete. Now describe Stephen Hawking. I don't think metabolism and brain power are related. --DHeyward (talk) 05:54, 14 May 2015 (UTC)

For all we know Hawking may well have a faster metabolic rate compared to people of similar size who suffer from the same condition. And if we compare him to an able body physicist then one has to take into account that with his muscles wasted away, there is more room for his body to power his brain. Count Iblis (talk) 14:33, 14 May 2015 (UTC)

For all we know, Stephen Hawking may actually be Davros in disguise, and have an entirely non-human physiology. When it gets to the point of arguing over hypotheticals we're lost. Your basal metabolic rate accounts for 70% of the calories you burn. Doubling that rate without huge and rather quickly fatal consequences is just not within the realm of physiological possibility.

A much simpler idea might be something like improved blood flow to the brain. I noticed after having my neck adjusted by the chiropractor that the lights would seemingly become brighter. I suspect that rather than him fiddling with the switch, the increase in oxygen flow through the blood vessels in effect turned up the brightness in my head. The opposite effect, dementia caused by restricted blood flow is well documented. μηδείς (talk) 17:54, 15 May 2015 (UTC)

The more stuff you burn the more waste you generate. I don't understand the strive for higher metabolism. It seems to be a shortcut from a healthy diet. 69.22.242.15 (talk) 09:45, 17 May 2015 (UTC)

Self-driving cars and speed limits

How much is known about the higher-level strategies being pursued by self-driving car manufacturers regarding how aggressively to drive, and in particular, regarding speed limits?

Searching, I find news items where the self-driving car dutifully goes at exactly the limit while all other traffic funnels around it, [4] and also where Google's cars are programmed to go up to 10 miles over like real drivers "if necessary" to keep up with traffic. [5][6][7] However, I don't know how many of Google's 1.7 million miles [8] were logged at over the speed limit - indeed, I wonder if they can even say without being subpoenaed for statistics that could provide greedy states with millions in fines?

One question is whether the puny human held in the vehicle has any say over whether to violate a speed limit when there are no other cars around, as they might be wont to do in less effectively administered cars.

Another is what the automated car should do when (as too often happens) other drivers start whipping past at 15, 20, 25 miles faster than the speed limit. In particular, if speeding further is not an option, I am curious whether the designers have built in enforcement capabilities. I remember one of the automated car designers had been quoted saying that non-automated driving would eventually be illegal - what better way to start than by having their excellent 360 degree video capabilities collect images of other cars driving past, coupled with both speedometer and GPS data from the automated car, uploaded directly to police each and every time a vehicle passes the automated car? Continuing with reports whenever someone fails to signal, fails to stop behind the stop sign, etc., until at last every last pathetic human admits that it is cheaper to take a very subprime loan on a new Google car than to try to face down an unending series of citations until their license is revoked anyway. Does the design of these cars have support for automated police reports; has the legal framework been laid for automated prosecutions? (Since unlike with an actual policeman with a radar gun, there would be no witness needing to show up in court) I feel like someone must have looked into such predictable consequences - have they? — Preceding unsigned comment added by Wnt (talk • contribs) 20:28, 11 May 2015‎

Our traffic laws currently rely on spotty enforcement. If you got a ticket every time you went 1 mile over the limit, changed lanes without using your blinker (I got a ticket for this !), or failed to come to a full and complete stop at a stop sign, you would lose your license after a single drive. Increased enforcement, by any means, would make this situation untenable. There are even situations where it's impossible to follow the law. I've seen stop signs at the entrance ramp to a freeway, with a minimum speed of 45 mph. How can you go from stopped to 45 MPH instantly ? If defies the laws of physics. So, enhanced enforcement would lead to calls for more reasonable traffic laws, which, in the long term, might be best. StuRat (talk) 20:46, 11 May 2015 (UTC)

While there is no problem with requiring automated cars or humans to use turn signals or stop at stop signs, there may be an actual conflict between laws requiring you to do no more than the speed limit and laws requiring you not to obstruct traffic. Rmhermen (talk) 21:09, 11 May 2015 (UTC)

Requiring cars to come to a full and complete stop at each stop sign might slow things down more than you think. A large portion of the time I arrive at a stop sign, I have a choice of slowing to under 5 MPH, then going right away, or coming to a complete stop, then waiting for a huge line of cars to pass. I drive quite conservatively, so I certainly don't do anything risky. I also think it's harder on cars to come to a complete stop, when they don't need to. Almost all stop signs should really be yield signs. The only exception I can think of is a case where you can't see if there is traffic on the cross street until you are parked there. Again, higher enforcement levels may force them to make more reasonable laws, eventually, once it become clear how painful it is to follow the laws by the book. StuRat (talk) 02:45, 13 May 2015 (UTC)

In many jurisdictions the right to confront your accuser becomes a right to question the technicians / manufacturers responsible for any automated evidence brought against you. If you choose to fight the automated ticket, then they have to produce a responsible human party in a timely manner to explain the operation of the machine and testify to the reliability of its operation or the evidence must be thrown out. I can't imagine Google would really want to tie up the lives of their technicians testifying in traffic court, even if they could make machines capable of reporting violations. Dragons flight (talk) 21:05, 11 May 2015 (UTC)

I can easily see roads, at least highways, being limited to self-driving cars once they're widely adopted. Once that happens, the cars can use their own sensors and talk to each other to determine the best safe speed, rendering the concept of a fixed speed limit obsolete.

As for automatic enforcement, there's some precedent with speed cameras. From what I've seen, these are usually handled as civil violations rather than misdemeanors, which allows them to get around some of the requirements for being able to contest the ticket with appeals outside the normal court system. But that generally requires each city and town to pass ordinances to implement it. But I imagine such a system would be more likely to lead to violent retaliation against self-driving car owners rather than faster adoption. Mr.Z-man 21:13, 11 May 2015 (UTC)

This is not directly relevant to enforcement with self-driving cars, but in some jurisdictions traffic offenses are already civil infractions rather than misdemeanors. In the District of Columbia, for instance, the only traffic offenses that are misdemeanors are reckless driving and driving under the influence, because those are the only offenses for which the penalty includes imprisonment, requiring criminal trial. Other traffic offenses are subject to administrative law, as are a variety of other matters formerly dealt with as misdemeanors, such as housing code violations. Robert McClenon (talk) 21:37, 11 May 2015 (UTC)

A cynic might view this as a way to enhance their "revenue stream" without leaving any recourse to victims. StuRat (talk) 22:47, 11 May 2015 (UTC)

It doesn't have that effect. It simplifies the procedure for challenging a ticket under administrative law. The trial is conducted by mail. It doesn't require a personal appearance in court. Changing traffic violations from criminal offenses to administrative infractions reduces costs both for the government and for any defendants. Robert McClenon (talk) 15:05, 12 May 2015 (UTC)

Yea, i tried that once. I explained in great detail that there was a blockage on the road and I, along with everyone else, had to pass it on the shoulder. I was found guilty or "responsible", and there was no response to my argument. I seriously doubt if anyone even read it. Made it quite clear to me that their purpose was to steal my money, not make the roads safer. Them collecting fines is a massive conflict of interest, so this result is to be expected. StuRat (talk) 02:32, 13 May 2015 (UTC)

I have two comments. First, the district judge or associate judge in a misdemeanor trial may be just as biased in favor of the Department of Motor Vehicles or police department as the administrative law judge in an administrative trial. Second, having not read your letter to the administrative judge, but having read your summary, if you said in your letter that their real purpose was to steal your money, that might very well have distracted the administrative judge from the merits of your defense. Robert McClenon (talk) 17:26, 13 May 2015 (UTC)

No, I'm smart enough not to say that to a judge (not that one ever read my letter). You're probably right about them all being biased against citizens and towards believing whatever police say, sometimes even in the face of contradictory video. Only a trial by jury has the potential to bring in somebody neutral (I say potential, because there are ways to get a biased jury too, like a change of venue, as in the Rodney King case) . StuRat (talk) 18:10, 15 May 2015 (UTC)

A car that automatically snitches on other drivers sounds like a good way to wind up with four flat tires. Or worse. 74.113.53.42 (talk) 21:53, 11 May 2015 (UTC)

And a court date, with video evidence. True, maybe you can hide behind a burka (an act already becoming illegal itself in many places, supposedly for anti-Islamic reasons). But if Google updates ten million cars to start snitching one day, without warning, how much difference will a few burka-clad fanatics really make? Wnt (talk) 23:51, 11 May 2015 (UTC)

I'd bet good money that these cars have a tunable parameter file someplace that contains properties like "OBEY_SPEED_LIMITS" and so forth. The car's software can presumably do whatever the manufacturer cares for it to do in this regard. In all likelyhood, the laws regarding speed limits would have to be changed during the decades where we switch from human-driven to self-driving cars - and it would make good sense to pass laws allowing them to drive maybe 5mph to 10mph over the posted speed limit.

So it's likely that the answer to this question will not be known until appropriate laws appear - and even then, it'll probably vary from place to place.

One important thing to know though - you may think that traffic is moving quite a bit faster than the speed limit - but most cars have deliberate speedometer error...their speedometers set in the factory to produce a higher speed reading than the car is actually travelling! So in many cases, people who think they are driving 5mph over the limit may actually be driving at or below the limit.  : It used to be that you could use a separate GPS unit to measure the exact speed - but those are also typically mal-adjusted. If you really care - find a nice flat, empty freeway, set your cruise control to 60mph and time how long it takes to pass a few mile markers. At that speed, they should be 60 seconds apart...the actual time is likely to be between 63 and 66 seconds.

SteveBaker (talk) 01:02, 12 May 2015 (UTC)

Um, what? Really? You think a hand-held hiker's GPS would have a deliberate speed error just to mislead drivers? How did you find this out? --Trovatore (talk) 01:30, 12 May 2015 (UTC)

Nonsense. Sorry steve, i've got 2 GPS and always check my cars against them, and have also checked against measured km. There frankly isn't a whole lot else to do between Melbourne and Sydney. Cars in general read fast (although modern cars are typically only 2% fast), GPS is spot onGreglocock (talk) 02:37, 12 May 2015 (UTC)

Yes, the same applies here in the UK. Car speedometers can read up to 10% high (closer to true speed for newer cars) but GPS units are exactly correct when measured against against motorway mile markers. Dbfirs 06:53, 12 May 2015 (UTC)

A cheap GPS might give a poor answer as well. The most typical problem is position jitter. Rather than tracking a straight-line course, at any given moment the GPS might be off +/- 10s of meters in some random direction giving a zig-zag line that appears longer than true path. Depending on the errors, the apparent distance and speed might be slightly overestimated. Smarter GPS units automatically smooth out those errors to give a better estimate of the track. Dragons flight (talk) 07:20, 12 May 2015 (UTC)

Yes, good point! The GPSr units that Greglocock and I use must be the smarter variety. Dbfirs 07:31, 12 May 2015 (UTC)

I should think the software could compensate for that, say by fitting a curve through the data points. Of course, that might underestimate the distance, if you really do drive in a zigzag pattern. StuRat (talk) 10:21, 12 May 2015 (UTC)

GPS speed readings are not computed by taking the distance between updates and dividing by the update interval. I mean, for all I know that may be one input to the computation, seems likely. But they also take into account the Doppler shift. I would imagine they use some algorithm that takes both things into account to reduce error. --Trovatore (talk) 18:02, 12 May 2015 (UTC)

Right. Two things to know:

1. The analysis the GPS receiver is doing on the radio signals it's receiving from the satellites gives it direct estimates of both its position and velocity.
2. It's typically feeding both its position and velocity estimates (along with their uncertainties) into a Kalman filter, whose job it is to sort of play these estimates off against each other, and over time, to get the best integrated estimate of your position and velocity, taking all the uncertainties into effect. The position estimate influences the velocity estimate, but also vice versa. It's possible for the receiver to end up being sure of one but not the other.

But, indeed, it is not doing anything nearly so straightforward as just taking the distance between two positon fixes and dividing by time. —Steve Summit (talk) 21:53, 12 May 2015 (UTC)

Yes, a smart GPS could do that. A car navigational GPS might also include an integrated compass and accelerometer for a further validating stream of information, and also use the incorporated map data to assume that cars are on the road surface most of the time and potentially ignore random errors that would place the driver slightly off of the road. That said, just because some GPS units are smart doesn't mean that they all are. The wrist mounted GPS units sold to runners and hikers to track their exercise often ignore the GPS velocity data. That's primarily because the velocity of the unit on a swinging arm is a poor indication of the person's actual rate of travel. In most such units, velocity really is determined solely from position and time, though often with some intelligent smoothing applied. The GPS services in smartphones tend to be something of middle ground. There is no reason, in principle, they couldn't incorporate a lot of the smart technology, but a manufacturer very concerned with power usage and chip set size sometimes chooses to use only very basic and dumb GPS receivers that provide little in the way of intelligence or error correction. Dragons flight (talk) 17:02, 13 May 2015 (UTC)

@SteveBaker: That wasn't intended as saying you're wrong. If GPS's do this sort of thing, I'd really like to know.

I mean, I certainly get why auto speedometers would generally be set to avoid reading low, so that no one sues for thinking he was not going as fast as he was and then getting a ticket or getting in a crash. The games I've seen at least one speedometer play, like reading accurately up to 70 mph or so and then starting to err more and more on the high side, are a little more annoying and I don't agree with it but I would sort of understand it. But if GPS makers are going along with the gag and falsifying a general-purpose speed measurement to keep you from finding out what your speedometer is doing — well, that would piss me off. --Trovatore (talk) 17:40, 12 May 2015 (UTC)

My aunt once got a speeding ticket and insisted she wasn't speeding. My uncle (a very skilled mechanic) eventually figured out that the manufacturer had installed a speedometer gear for the wrong axle ratio. Short Brigade Harvester Boris (talk) 00:10, 13 May 2015 (UTC)

In the UK legal action isn't normally taken (note "normally") unless you are going quite a bit over the speed limit (10% + 2 mph).[9] However, maybe this allowance wouldn't be given for cars programmed to travel above the limit. Thincat (talk) 07:47, 12 May 2015 (UTC)

Sebastian Thrun, who has been the creative force behind much of the new era of autonomous vehicles, wrote Online Speed Adaptation using Supervised Learning for High-Speed, Off-Road Autonomous Driving. "The mobile robotics community has traditionally addressed motion planning and navigation in terms of steering decisions. However, selecting the best speed is also important..." Thrun also wrote extensively on the use of statistical methods for robot decision making. His stuff is worth reading.

The point is, robotic cars that use modern technology don't have a "speed limit." They have a complex artificial intelligence software that governs their decision-making for high-level tasks, like path planning, and for low level tasks, like control systems for the motor. SteveBaker's idea of a tunable knob is a step in the right direction - and perhaps there even exists some source-code line that does set a hard limit - but it's a lot more probable that the robot brain is using a statistical method to select its speed - and it makes the decision using some form of a fancy weighted average of many different abstract inputs, including sensors, information about traffic rules, vehicle dynamics, and programmer-defined heuristics. Although Google doesn't publish much about its current algorithms, you can read about the history: there are many speeds of importance, including a "never exceed," a "shock threshold" speed, and perceived (estimated) speeds from multiple different types of sensors (cameras, rotary encoders, GPS and radio navigation systems, remote sensors, vehicle information systems, ...)

Regarding the machine learning and decision-making method for speed limits: (excerpted from the paper I linked above): "Notice that the speed limits are often much greater than the safe speed the human (and the algorithm) selects. This demonstrates the importance of our technique over blindly following the provided speed limits. Notice also, however, both the human and the algorithm at times exceed the speed limit. This is not allowed on the actual robot ... " (at least, not during operation of the Stanley robot at the DARPA Challenge in 2005).

Nimur (talk) 14:21, 12 May 2015 (UTC)

Yes, for sure there would be more to it than simple "Never go faster than X mph on a road with a Y mph speed limit" rule - you'd want the car to be able (for example) to briefly accelerate over the limit in order to avoid certain death for the occupants, for example. But whatever feeds into the "Ideal-cruising-speed" algorithm would obviously be written with the possibility to modify the legal speed limit before feeding it into that calculation.

In a world where there are ONLY self-driving cars, the entire concept of a speed limit seems kinda silly. We have speed limits in order to limit fuel consumption (that was the original reason for the 55mph limits in the USA) - and to provide some degree of safety for other road users. But if the car's robotics are factoring in things like gas consumption (or battery power) - and can actually know the safe driving speed given all of the other variables - why would we even need a speed limit? Rather, you'd need some laws about maximum energy consumption and accident probabilities - and provide a certification for cars that meet the desired criteria, rather than a blanket speed limit which is a messy legacy from human drivers and poor "catch-all" assumptions about the optimum speed-versus-energy which was only ever true for the "average" car.

A robotic car might very well want to factor in the expected life of it's batteries and the distance to the destination in figuring out the optimum speed for a given set of traffic densities, etc.

You also have the idea of robotic cars that are designed to deliberately tail-gate other robotic cars in order to minimise wind-resistance - and in such cases, there is an enormous win to having them all choose to go at precisely the same speed.

Self-driving cars will be VERY different from human-driven vehicles...they don't have to behave the same way at all. If/when they become the only kinds of vehicles allowed on the road, then who knows where we'll end up? For example - why do we need traffic lights or STOP signs? If the cars can reliably and fairly negotiate who gets to go though the junction and when, then the traffic can make small speed changes as they approach the intersection and criss-cross in an exceedingly alarming manner to optimise speeds through a junction. Stuff like that is easy to do when there are no humans around to mess things up...and a radio network between the cars is a much better solution than having to have in-car cameras that look at a glowing light to tell them when they are or are not allowed to go from a dumb cyclic timer.

If you were able to design how cars drove and how road systems are designed from the ground up - without all of that legacy crap to deal with, you could do the job VASTLY more easily.

SteveBaker (talk) 18:38, 12 May 2015 (UTC)

But a major reason for speed limits is the safety benefit of vehicles all going at a more or less uniform speed. If each vehicle went at its own optimum speed you'd end up with widely varying speed according to the capability of each vehicle. Traffic engineers observe that the likelihood of a crash happening is significantly higher if vehicles are traveling at speeds faster or slower than the mean speed of traffic. Short Brigade Harvester Boris (talk) 00:19, 13 May 2015 (UTC)

True - but think outside the box here. With smart cars that can talk to each other at high speeds and negotiate fairly, they could dynamically negotiate a common speed that all of the nearby 'pack' of cars could be happy with. On a big open freeway at 4am, a group of high powered sports cars who happen to be close to each other and far from other vehicles could agree on a common speed far higher than on that same freeway when it's packed with big-rigs and rusty old clunkers. If driving at a common speed makes sense, then that's something that's better arranged by letting the robotic cars negotiate than it is by legislating for the worst possible case. SteveBaker (talk) 03:56, 13 May 2015 (UTC)

Traffic conditions are not necessarily the only factor to consider. Speeds may be reduced do to the potential for dangerous road conditions other than traffic, e.g. children playing, large animals, rockfalls, uneven surfaces, icing, etc. Or even for reasons like noise abatement / smog control that have nothing to do with road conditions at all. I agree that a fully automated highway could do a better job of tuning the flow of traffic and many speed limits might be relaxed, but I doubt that the government would stop setting speed limits entirely. Dragons flight (talk) 17:16, 13 May 2015 (UTC)

Agreed. Actually I suspect this suggestion is interesting from a sociological point of view. For many years it was well known that speed limits in the Western U.S., for example, were much too slow for the flat terrain and low-traffic conditions, and were motivated by fuel efficiency concerns. An individual might decide to go 100 mph on such a road, and could do so safely, but no proof of safety would help him in court. But with self-driving cars, there's a sort of apocalypse: on one hand, you might decide that machines have a higher social status than humans, are more trustworthy, or are effectively under the control of corporations with more political pull, and say that they can violate the rules that humans cannot, even when the human could prove he was justified. Alternatively, you might limit the machines to the speed limit, but then the people using them may be severely dissatisfied with their performance. The obvious middle ground would be that the rider has the option to set the machine to speed however much he wants it to - this is the case with cruise control or a heavy foot, after all - yet it seems to conflict with widely held social norms that expect companies to be the overseers of all their customers might do with their products. Wnt (talk) 11:53, 14 May 2015 (UTC)

You're missing the possibility that I mentioned, that better enforcement of laws ultimately will lead citizens to demand more just laws. While nobody much cared if the speed limit was 55 MPH when they were free to ignore it, once everyone is forced to comply, there will be a huge outcry. StuRat (talk) 18:20, 15 May 2015 (UTC)