[...] The 54 [Felton line], brought to a halt by an autonomous vehicle belonging to Alphabet's Waymo, isn't the only bus that's run into trouble with San Francisco's growing crowd of driverless vehicles. Bus and train surveillance videos obtained by WIRED through public records requests show a litany of incidents since September in which anxiety and confusion stirred up by driverless cars has spilled onto the streets of the US city that has become the epicenter for testing them.
As the incidents stack up, the companies behind the autonomous vehicles, such as Waymo and General Motors' Cruise, want to add more robotaxis to San Francisco's streets, cover more territory, and run at all hours. Waymo and Cruise say they learn from every incident. Each has logged over 1 million driverless miles and say their cars are safe enough to keep powering forward. But expansions are subject to approval from California state regulators, which have been pressed by San Francisco officials for years to restrict autonomous vehicles until issues subside.
Driverless cars have completed thousands of journeys in San Francisco—taking people to work, to school, and to and from dates. They have also proven to be a glitchy nuisance, snarling traffic and creeping into hazardous terrain such as construction zones and downedpower lines. Autonomous cars in San Francisco made 92 unplanned stops between May and December 2022—88 percent of them on streets with transit service, according to city transportation authorities, who collected the data from social media reports, 911 calls, and other sources, because companies aren't required to report all the breakdowns.
The records obtained by WIRED are more focused. They follow a previously unreported directive to staff of the San Francisco Municipal Transportation Agency handed down last October to improve record keeping of incidents involving autonomous vehicles. Muni, as the agency is known, standardized the term "driverless car" when staff report "near-misses, collisions or other incidents resulting in transit delay," according to the directive. Agency logs show 12 "driverless" reports from September 2022 through March 8, 2023, though Muni video was provided for only eight of these cases. Overall, the incidents resulted in at least 83 minutes of direct delays for Muni riders, records show.
That data likely doesn't reflect the true scale of the problem. Muni staff don't follow every directive to the letter, and a single delay can slow other lines, worsening the blow. Buses and trains cannot weave around blockages as easily as pedestrians, other motorists, and cyclists, saddling transit-dependent travelers with some of the biggest headaches caused by errant driverless cars, according to transit advocates.
That left the Muni driver in a bind. "I can't move the bus," the driver said to one of two riders on board. "The car is automatic driving." The driver radioed managers and doffed their cap: "Whoosh ... Half hour, one hour. I don't know. Nothing to do." Thirty-eight stops and about five miles remained ahead for the 54. The driver, looking out at the Waymo, expressed disappointment: "This one not smart yet. Not smart. Not good."
Waymo's Karp says one of the company's roadside assistance crews arrived within 11 minutes of being dispatched to drive the SUV, clearing the blockage about 15 minutes after it began. Karp declined to elaborate on why the remote responder's guidance failed but said engineers have since introduced an unspecified change that allows addressing "these rare situations faster and with more flexibility."
The Transport Workers Union, which represents Muni train and bus drivers, deferred comment for this story to Muni. The agency declined to make drivers described in this story available for comment. But Tumlin, the Muni director, says San Francisco's transit workers are frustrated. "When you encounter a vehicle with no human on board, it is dispiriting and disempowering," he says. "There's no one there to communicate with at all."
(Score: 4, Insightful) by looorg on Wednesday April 12, @11:31AM (13 children)
Perhaps buses and such should just have a plough, wedge or cowcatcher installed so it can push other cars (driverless or not) off the road. Clearly they are in the way. Might or size, and physics, makes and give(s) right of way.
(Score: 5, Interesting) by canopic jug on Wednesday April 12, @11:47AM (4 children)
Either way, there has to be some incentive to get the unmanned vehicles off the road. Otherwise, rather than pay for parking, people will just send the vehicle out to drive around aimlessly in the city while they sit at work for the day.
Money is not free speech. Elections should not be auctions.
(Score: 0) by Anonymous Coward on Wednesday April 12, @03:50PM (1 child)
That's still cheaper than paying for parking.
(Score: 3, Interesting) by canopic jug on Wednesday April 12, @04:24PM
That's still cheaper than paying for parking.
For the individual, maybe. For society, definitely not. The incentives to park the cars when not in use has to be quite strong as well as independent of money.
Money is not free speech. Elections should not be auctions.
(Score: 1, Insightful) by Anonymous Coward on Thursday April 13, @01:38AM (1 child)
(Score: 3, Funny) by kazzie on Thursday April 13, @07:16AM
Any cases would be obstructed in the courts. By parking cars there.
(Score: 0) by Anonymous Coward on Wednesday April 12, @02:02PM (1 child)
Add flame-throwers and grenade launchers for those "hard to clear" clogs and I'm ready to invest in your idea to get it off the ground!
(Score: 0) by Anonymous Coward on Wednesday April 12, @08:00PM
Or Javelin missiles.
Here's a recently aired (11 April 2023) documentary [imdb.com] about some civic-minded folks who want to use those to clear up a traffic jam in the downtown area of a small city.
Sadly, government overreach stops them from doing so. it's very sad. Defund the FBI! [reuters.com].
(Score: 2) by mcgrew on Wednesday April 12, @02:19PM (5 children)
I don't know about your shithole country, but here in the US, the laws state that right of way is ALWAYS given and NEVER taken. You sound like someone who should never be allowed to drive a motor vehicle. WTF is wrong with you?
Carbon, The only element in the known universe to ever gain sentience
(Score: 4, Interesting) by JoeMerchant on Wednesday April 12, @02:27PM (4 children)
But when a vehicle has no live occupants, not even a driver, does it still have the same "rights of way" as a person?
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 2) by krishnoid on Wednesday April 12, @04:27PM
Uh no, because the person can flip off the vehicle. That overrides everything until they equip the vehicles with an externally-facing heads-up display on the windshield that can display various rude hand signals (and loudspeakers that can yell curse words) based on its ethnic/national profile of the person.
Seriously, even when computers are monitoring other computers they all still drive like jerks. I guess bad behavior in municipal traffic is more of a universal truth than the impending dominance of artificial intelligence.
(Score: 2) by mcgrew on Thursday April 13, @09:50PM (2 children)
No. Just as if it's illegally parked, you tow it away. You don't "push it off the road", what's wrong with people?
Carbon, The only element in the known universe to ever gain sentience
(Score: 3, Insightful) by JoeMerchant on Thursday April 13, @11:46PM (1 child)
Not advocating pushing anything off the road, but it shouldn't have any more rights than an improperly parked / abandoned vehicle, particularly if there's no person inside.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 2) by mcgrew on Monday April 17, @06:25PM
Nobody said it should have more rights. If anything, it should have no rights at all like any other mindless machine, although its owner has rights and responsibilities.
Carbon, The only element in the known universe to ever gain sentience
(Score: 2) by wisnoskij on Wednesday April 12, @12:44PM (11 children)
What is that even supposed to mean? I don't care how much maintenance the cars need, planned or not. I care how many people they have run over, and how much they have slowed or sped up traffic.
(Score: 4, Interesting) by Nuke on Wednesday April 12, @01:32PM (8 children)
I'll translate : 92 occurences of SD vehicles stopping and remaining stopped in places they should not and consequently holding up other traffic.
I guess more of the granular detail will not be generally available, but I know that in the centre of my city (in Europe) even a short hold up (such as cause by a geriatric pedestrian crossing the road slowly) can cause a domino effect creating a traffic jam that lasts for a long time over a wide area.
There was an infamous afternoon a few years ago when a single broken down Skoda at a cross roads caused an enduring gridlock across the entire city - at 5 pm it took me an hour just to get out of my company car park and another two hours to go another mile to the outskirts and be free of it.
(Score: 5, Insightful) by JoeMerchant on Wednesday April 12, @02:31PM (7 children)
The stats that really matter are: how much X per Y passenger miles as compared to human driven vehicles?
If these driverless cars haul people around for 50,000 miles between breakdowns, and they "back up traffic" the same or less than human driven cars when they breakdown, then I don't care if there are 1000 of these events - that's a decent record.
Same for accidents with property damage, injuries, or deaths. If driverless cars in the US are killing 2 people per hour, that could be a significant improvement over meatbag pilots.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 4, Insightful) by ikanreed on Wednesday April 12, @05:05PM (6 children)
Yeah, and when we were first talking about driverless cars, purely conceptually, arguments like yours swayed me. Now that we see that companies like Tesla engage in cynical "turn off the autopilot just before a collision" type shit to game the accident numbers they report, I'm no longer sympathetic to such arguments. This shit needs to go.
(Score: 2) by JoeMerchant on Wednesday April 12, @05:19PM (5 children)
>companies like Tesla engage in cynical "turn off the autopilot just before a collision" type shit to game the accident numbers they report, I'm no longer sympathetic to such arguments.
Yep, that's no auto-driver at all as far as I'm concerned.
The new cars I have rented with driver assist, it's a load of crap - basically binging you when it thinks you're drifting out of the lane, sometimes nudging the wheel, but never taking actual responsibility, and IME that's a good thing because I have found several road situations where the lane assist will nudge for a swerve based on meaningless color patterns on the pavement. I think I've been auto-braked for no good reason once or twice too.
One thing I do appreciate in the "advanced" driver assist modes is: adaptive cruise control with selectable following distance. Again, it needs meatbag correction some times, but that is actually a control loop I appreciate the help with - particularly for setting long following distances on two lane roads.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 3, Interesting) by fliptop on Wednesday April 12, @05:32PM (4 children)
A feature like that would be awesome for drafting behind a semi. I do that often, travel in the "slipstream" to squeeze out extra gas mileage. How close does it allow you to get?
To be oneself, and unafraid whether right or wrong, is more admirable than the easy cowardice of surrender to conformity
(Score: 2) by JoeMerchant on Wednesday April 12, @06:03PM
The ones I have used (VW most recently, I think) have had 4 following settings, which seem to roughly correspond to number of seconds behind the vehicle in front. One second following distance is perhaps a little close, but not tailgating or drafting yet. The longest setting might have been more like 5 seconds, which is pretty nice for 2 lane driving through the countryside.
Україна досі не є частиною Росії Слава Україні🌻 https://news.stanford.edu/2023/02/17/will-russia-ukraine-war-end
(Score: 2, Interesting) by Anonymous Coward on Wednesday April 12, @06:25PM (2 children)
> drafting behind a semi
In my youth I did this occasionally, once we did it with an early VW Beetle (1192cc engine). The draft was strong enough to overspeed the poor little engine and when the semi sped up on a downgrade we blew the motor.
However, once you see a re-tread tire come apart and spit out that heavy slab of rubber, you will probably stop drafting trucks. Trailer tires are frequently re-treads, many truck tires are designed for this and retreading several times is common. But sometimes the new tread doesn't bond completely to the reused tire.
A few times I remember the semi driver swerving gently back and forth when I got up into drafting distance, a signal to back off. That driver may have known something that I didn't (for example, his truck had questionable tires?)
(Score: 3, Interesting) by Nuke on Wednesday April 12, @07:24PM (1 child)
No, he was just pissed off with you riding on his coat tails. Truck drivers also believe (rightly or wrongly) that their own gas consumption is increased.
(Score: 4, Informative) by Anonymous Coward on Wednesday April 12, @08:48PM
Wrongly. It actually decreases their diesel consumption. The low pressure area behind a truck is a major drag. Having a car drafting increases the pressure and reduces the drag slightly.
(Score: 2, Touché) by mcgrew on Wednesday April 12, @02:21PM (1 child)
You know, reading the fucking article stops a whole lot of ignorant, uninformed, STUPID questions like yours. There's a link, use it!
Carbon, The only element in the known universe to ever gain sentience
(Score: 0) by Anonymous Coward on Thursday April 13, @07:44AM
RTFA? U must be new here.
(Score: 3, Funny) by Isia on Wednesday April 12, @12:45PM
I hope Tesla recognizes the level of perfection that people expect from Robo-Taxis and don't hit the street with a less than 99.99999% perfect Autopilot.
(Score: 2) by DannyB on Wednesday April 12, @02:28PM (1 child)
One thing that self driving cars could be maid to do is to cooperate with each other so that they all experience the most efficient travel possible.
Cooperating while driving is considered a major offense amongst humans. That is why humans will never engage in this abhorrent cooperation behavior.
How often should I have my memory checked? I used to know but...
(Score: 3, Interesting) by Reziac on Thursday April 13, @03:01AM
I used to live out in the desert north of Los Angeles. And in our midsized town, you could count on driver cooperation -- four way stops, yields, turns, lane changes. Lots of "After you, my dear Alphonse."
Then we started getting an influx of New Bedroom Community from L.A., and over the space of a few years every interaction on the road became competitive rather than cooperative.
Similarly, down in L.A. -- in old downtown, no matter how crowded the road, if you put on your signal sooner or later someone would let you in; downtown had been there a hundred years and wasn't going anywhere, so no hurry no worry. But over in West L.A., New Money Yuppieville -- it was all cut off everyone else, ALL the time, like they were terrified of being last in line.
Really a stark difference in driving philosophy.
Now I'm back in the Northern Wastes, and it's nice to experience cooperative driving again.
And there is no Alkibiades to come back and save us from ourselves.
(Score: 3, Insightful) by bryan on Wednesday April 12, @02:53PM (1 child)
Err, does this mean the bus only had 2 passengers?
(Score: 2) by kazzie on Thursday April 13, @07:28AM
Maybe the rest had already realised what was going on, and disembarked?
Mind you, even in rush hour, buses going against the flow (going to collect passengers) can be pretty empty.
(Score: 1, Informative) by SomeRandomGeek on Wednesday April 12, @04:33PM (11 children)
So, the bus is traveling down a two way street that is only one lane wide. It comes across a self driving car coming the other way down the street. Impasse! The bus driver is not allowed to back up without permission from his supervisor. So, the self driving car must back up. The bus driver initiates his standard procedure of honking and screaming, but it doesn't seem to work. The self driving car refuses to back up. The self driving car and the bus both radio for help. Fifteen minutes later, a human arrives to coach the self driving car out of this mess, and the car finally backs up. Meanwhile, the bus driver is still honking and screaming.
Clearly, all the blame lies with the self driving car.
(Score: 3, Touché) by Anonymous Coward on Wednesday April 12, @06:33PM
> Clearly, all the blame lies with the self driving car.
Yes, all the blame lies with the self driving car (although I detect satire in your tone).
A person driving that car would see the bus coming down that congested city street and move out of the lane somehow, to make room for the bus to pass. For example, the person might pull into a "no parking" spot next to the curb, temporarily.
(Score: 1, Informative) by Anonymous Coward on Wednesday April 12, @07:21PM (3 children)
That sounds like bullshit. There are so many situations where that might be necessary in an urban environment that your (unsupported by any documentation of such "rules") assertion appears to be ridiculous on its face.
I suppose I could be wrong, but I doubt it.
(Score: 2) by SomeRandomGeek on Wednesday April 12, @08:09PM (2 children)
Not my assertion. It was in the article.
(Score: 1, Funny) by Anonymous Coward on Wednesday April 12, @08:35PM
In that case, I question the author of TFA for that assertion (assuming they don't provide supporting documentation for that) rather than you. My apologies.
Although I will take you to task for reading TFA. That's just wrong. And I urge you to stop doing such evil things! ;)
(Score: 2) by Reziac on Thursday April 13, @03:05AM
Guessing because of the hazards of blind spots, and also the problem of finding bus drivers who are actually competent to back up such a large awkward vehicle with an ass that pivots way beyond the rear axle.
And there is no Alkibiades to come back and save us from ourselves.
(Score: 4, Touché) by Nuke on Wednesday April 12, @07:27PM (5 children)
Yes, it does. It is common sense not to require the larger and more cumbersome vehicle to reverse for a distance.
(Score: 2) by SomeRandomGeek on Wednesday April 12, @08:25PM (4 children)
I see five different failures:
1. The city of San Francisco -- For allowing two way traffic on a road only wide enough for one, or alternatively, for allowing parking on both sides of the street so there is only one lane left for traffic.
2. The transportation authority -- For putting a bus route down a street that is clearly not wide enough for it.
3. Waymo -- For allowing the car to go down that street.
4. The car -- For not backing up.
5. The bus -- For not backing up.
I agree with you that of those five failures, the least blame should be assigned to the bus for not backing up. But I think the second least blame should be assigned to the car for not backing up. It is not reasonable to expect either the car or the bus to deal with a two way road that is one lane wide. And backing up through an intersection (which is what you are expecting the car to do) is just begging to get someone killed.
(Score: 2, Informative) by Anonymous Coward on Wednesday April 12, @09:36PM (2 children)
> It is not reasonable to expect either the car or the bus to deal with a two way road that is one lane wide.
Said the person that most likely has not driven in older cities (particularly in Europe) where the narrow streets and buildings on the street were built before cars...
(Score: 4, Funny) by hendrikboom on Thursday April 13, @12:53AM (1 child)
I once saw someone in a large American car try to turn a corner in Amsterdam from a narrow street to a narrow street beside a canal. It was a lot of going back and forth while managing to turn a bit more each time. They did succeed. The driver must have been good at parallel parking, too. But it took about ten minutes and drew a crowd of about ten or twenty locals. When they finally succeeded the crowd applauded.
It makes me think there should be a video game in which the player has to park implausible vehicles in implausible parking spots.
(Score: 2) by kazzie on Thursday April 13, @07:32AM
GTA-P (Get The Auto Parked)?
(Score: 0) by Anonymous Coward on Thursday April 13, @01:32AM
In the real world it's reasonable to expect drivers to deal with such stuff. A bunch of vehicles could have illegally parked for some reason (possibly even legit or maybe failed self driving vehicles) and caused it to effectively become a one lane road. Items might spill on the road. Stuff happens. That said even human drivers often don't deal well with two way one lane situations (think road rage and guns) HOWEVER in the actual scenario there were enough spots and spaces for the car to squeeze into which is what most human drivers in my area would have done.
Do buses in SF have the right of way in such cases? In some places street buses have the right of way.
(Score: 2, Informative) by Anonymous Coward on Wednesday April 12, @06:50PM (3 children)
From tfs,
> Each has logged over 1 million driverless miles and say their cars are safe enough to keep powering forward.
"How safe is safe enough for Autonomous Vehicles" was addressed by this recent online talk to a group of about 250 transportation professionals (transit authority, city planners, etc). https://www.youtube.com/watch?v=shp8co9oqYE [youtube.com]
The meat of the talk is about 20 minutes and the rest is discussion. The speaker is Phil Koopman from Carnegie Mellon, who has researched and published extensively on this topic, http://users.ece.cmu.edu/~koopman/ [cmu.edu]
I thought I had a good idea of the major problems before hearing the talk...and I went away realizing that there are many problems I hadn't even considered. The overall picture is much more complex than I originally thought. If you have 20 minutes and are interested in this topic you might be enlightened.
(Score: 2, Interesting) by Anonymous Coward on Thursday April 13, @03:02AM (1 child)
From the video it sure looks like the stuff is far from ready. In 51:00 they are not noticing construction workers just because they are the wrong color. And then the guy talks about what's the next color?
That's terrible.
Maybe you lack imagination, experience or don't realize how crap their shit is.
They're ready when they've fix all that crap and are at the level of figuring out how to handle human drivers who will actually bully AI vehicles because they know that they'll >90% stop/give way in certain cases.
(Score: 0) by Anonymous Coward on Thursday April 13, @11:23AM
Thanks for the followup and nice to see that at least one person took the time to watch the Phil Koopman video.
>> I went away realizing that there are many problems I hadn't even considered.
> Maybe you lack imagination, experience or don't realize how crap their shit is.
I think I'm doing OK on all three--for an interested spectator/amateur. Listening to a pro (Koopman studies self driving cars as his job) just added a bunch more details and filled in some blanks.
A future bowl of popcorn is reserved for the next round of self-driving permits in SF (and a few other cities as well) -- will the city let the fleets of self driving taxis grow?
(Score: 3, Informative) by hendrikboom on Thursday April 13, @07:29PM
That was a very interesting talk, including the question-and-answer session.
But I didn't need the first few minutes where they introduced the speaker.
-- hendrik
(Score: 2) by darkfeline on Wednesday April 12, @07:01PM (6 children)
And what are the stats for human drivers? I imagine they've caused more than 83 min of delays over seven months.
Join the SDF Public Access UNIX System today!
(Score: 2) by NotSanguine on Wednesday April 12, @07:52PM (5 children)
Very likely.
While traffic back ups are a pain for drivers (not that I really care, in fact it amuses me greatly), the more important statistic (IMHO) would be a comparison injury/death/property damage per vehicle mile driven of human drivers to that of the semi-autonomous* vehicles discussed here.
I posit that the above should be the key metric WRT to the efficacy/value of such semi-autonomous vehicles on our streets, roads and highways, with traffic tie-ups a secondary metric.
That said, the death/injury/damage per vehicle mile statistic needs to have significant data for both human-controlled and semi-autonomous vehicles. We have lots of data on the former (and it's ugly), but not enough of the latter.
Because the really important part here isn't the ability of SV rent-seekers not to have to pay taxi drivers, but rather to create safer roads and reduce the carnage seen every day [statista.com].
And we shouldn't let the perfect be the enemy of the good, either. If these semi-autonomous vehicles are even 15-20% safer than human-controlled vehicles (assuming we have enough data to make a valid comparison), we should encourage more widespread use.
And (apologies for my relentlessly optimistic view of technological advancement) that can improve significantly as issues with these semi-autonomous vehicles are identified and addressed.
At some point, (one would hope, at least) the numbers of such vehicles will increase enough to have a measurable positive effect on vehicle crashes and the associated deaths/injuries/property damage.
Until then, if your local/state governments actually give a rat's ass about those they're supposed to represent, you will have alternatives like decent public transportation, bike lanes and walkable areas, then you should use those modes of transport instead.
*I say semi-autonomous because if these vehicles were fully autonomous, unless there's actually physical damage to the vehicle that stops it from moving, a fully autonomous vehicle would be able to address whatever the issue might be and continue on. That's not the case (as TFS makes perfectly clear) with these vehicles.
No, no, you're not thinking; you're just being logical. --Niels Bohr
(Score: 2, Touché) by Anonymous Coward on Wednesday April 12, @09:43PM (4 children)
> more important statistic (IMHO) would be a comparison injury/death/property damage per vehicle mile driven
See post above,
pfft, a million miles is nothing
by Anonymous Coward [I'm that same AC] on Wednesday April 12, @02:50PM (#1301143).
While your comparison would be nice to make, there isn't enough data yet on the self driving vehicles to do it. Watch the linked video to see that it takes 10s or 100s of million miles to be statistically significant. For example, if you compare to the overall average for human drivers, that includes all the _drunk_ driving accidents. Surely you want your self driving vehicle to be at least as safe as the population of _sober_ drivers?
(Score: 2) by NotSanguine on Wednesday April 12, @09:50PM (3 children)
Did you not read this sentence (the second after the one you quoted) in my comment:
No matter. Literacy is overrated, isn't it? In any case, thanks for agreeing with me.
No, no, you're not thinking; you're just being logical. --Niels Bohr
(Score: 2) by NotSanguine on Wednesday April 12, @09:54PM (2 children)
Oops. Screwed up the link to the comment [soylentnews.org] to which you replied.
Sorry about that. I hope it didn't confuse you more than you already are.
No, no, you're not thinking; you're just being logical. --Niels Bohr
(Score: 0) by Anonymous Coward on Thursday April 13, @12:05AM (1 child)
No worries, I'm not confused. It's true that I didn't read all the way through your first post -- in true SN style I reacted as soon as I got to your line, the line that I quoted with a ">".
Going back to this comment,
> And what are the stats for human drivers? I imagine they've caused more than 83 min of delays over seven months.
I don't think it was well formed--In SF there must be several orders of magnitude more human drivers than driverless cars. Any comparison would need to know that ratio and proportion the delays appropriately, but getting the data on the humans might be difficult...
(Score: 3, Funny) by NotSanguine on Thursday April 13, @01:56AM
As I mentioned, literacy is, apparently, overrated.
But you came to the same (correct) conclusion I did -- that more data is needed.
Fair enough. Although my comment *quoted* that from a comment [soylentnews.org] by darkfeline.
And I agreed with them for exactly the reason you cite.
So I guess great (as well as mediocre -- but that certainly doesn't apply here! -- have I mentioned that my fabulousness is exceeded only by my modesty?) minds think alike, eh?
No, no, you're not thinking; you're just being logical. --Niels Bohr