Hacker News Reader

Comment by barrkel | Hacker News Reader

Something similar happens when riding motorcycles. At first you're cautious because you don't know how to evaluate the risks. Then you move into a peak of overconfidence where the usual risks are known; but this is the most dangerous period. It's only after a while at this level, perhaps after a few scares or accidents, that a new appreciation for unusual risks starts to come in. You start to become less optimistic, and consider what could go wrong, not just what's likely to go wrong. You start actively seeking reasons for the traffic around you to do the unexpected.

Mature evaluation of risks increases the weighting given to tail events, black swans etc., particularly if the outcome is severe. Teaching this perspective ought to be effective in keeping it in mind for the teacher too.

View on HN

My Dad - who used to drive an Ambulance then Fire Engines (for about 20 years) - taught me to drive and use something called "Defensive Driving Techniques". Where most people look at one or two cars ahead and hopefully the one behind, my Dad taught me to look at every car that I can see, using every gap, window, mirror etc to be aware of my surroundings. Look at people, bus stops, crossings, shops in towns etc.

He also suggested that rather than just looking at a car, you look at the position of the car, who's driving, the age, have I seen them drive weirdly yet? This way you can build up a much better mental model of how they're going to react to something. This has saved me numerous times where I've judged what a car will do just by how they're placed in the road, even if they look like they may do something else to a normal driver.

It's a technique that I've shown to my girlfriend and it's changed her driving (for the better) and what I'll be teaching my kids when the time comes.

That's how you drive in third world countries.

It's the right way to drive anywhere hundreds of barely-trained primates are distractedly operating heavy equipment at high speed...

That's a kind of brute force approach. It seems sub optimal but I don't know how to reason about it.

Should I say that observing every vehicle across the median on the oncoming carriageway is inefficient? Well I can't say that because occasionally some small cargo or even a wheel falls off and bounds across to the oncoming lane - something you definitely want to see coming.

Perhaps taking a zonal approach is as effective and certainly less taxing on the driver.

Scan each zone, some zones get scanned more frequently, and look for anything unusual.

Defensive driving is really just a specific application of situational awareness [0]. With practice and experience you'll go from just maintaining visibility and space for reaction time to actually remembering where drivers are that have been blocked from your view by other vehicles. You never really track every vehicle and pedestrian, but you do make note of particular ones and the rest get lumped into groups based on their behavior. You'll learn to anticipate the car that's going to make a left turn in front of you by taking your foot off the throttle earlier instead of braking at the last second.

[0] http://en.wikipedia.org/wiki/Situation_awareness

you're talking about chunking: at first you try to remember everything, then as you gain experience you learn you can abstract certain items away to a general model of them (chunk) without having to remember the specifics of each individual item.

"That's a kind of brute force approach. It seems sub optimal but I don't know how to reason about it."

Well, what else are you doing while driving?

For the most part, whatever your answer is to that question, stop it. (Reasonable exceptions apply; conversation with another in the car isn't too big a deal, of course.)

Now that you're just driving and not doing anything else, you might as well be active about it and be on the lookout for long-tail risks, since really long-tail risks are all you ever have to worry about. Driving is fairly safe and almost everything that ever happens bad is long-tail in the first place or we wouldn't drive.

Agree 100% with "if you are doing something else while driving, stop it". I have always thought the conversation is the distracting part, not the act of holding a phone to your head. Research shows that it's less distracting to talk to a passenger than a phone conversation: http://scienceblogs.com/cognitivedaily/2007/01/24/want-to-dr... (link in article). Still, I prefer to remain focused on the driving conditions.

I'm not sure I agree that "driving is fairly safe" though. We have mitigated much of the risk associated with driving with vehicle impact standards and road design. I still really feel that most drivers don't understand the magnitude of what they are doing - they become accustomed to the relative safety and start doing other things. It seems the more safety we "build into the system" the more risks people are willing to take. I'm not suggesting we take safety out of the system though!

A zone won't hit your car...another car will.

Some things need to be brute forced.

Like everything related to driving, it eventually becomes automatic.

It's not that you're consciously looking at each car and saying "Is this car going to be a problem? Is this one a problem?" Instead, you recognize patterns. If a car suddenly weaves out of its lane, it triggers a tinge of fear and you give that car a wide berth until you can pass him and leave him behind. If you see brake lights go on 10 cars ahead of you, you know that you're going to have to slow down, so get ready for it (oftentimes, I'll let up on the gas as soon as I see brake lights go on a quarter mile down the road, and by the time I get to the traffic, I'll be going their speed anyway). If you're in a car's blind spot, pay extra careful attention to turn signals or any deviation from his lane, and try to pull forward or behind so he can see you. If a car is clearly in a rush, it's a good bet he's going to weave in front of you, so hang back and make sure you have room to adjust if he does. If two cars try to enter the same spot in the same line (this is my personal biggest fear, they're both in each other's blind spot and each has reason to believe the lane they're switching into is clear), hang back, and honk if it's clear that they don't see each other.

It's about pattern recognition, knowing which situations are dangerous so you can be extra careful whenever it looks like one of those situations is happening.

I use that technique, and during a period in which I commuted, my meditations on cars taught me to use the same principles of thorough research and prediction to analyze stocks, which worked out well. It can also apply to understanding a room of people or a crowd. Those are all situations where overconfidence can be worse than overcaution.

I think defensive driving is a good way to introduce that kind of prudence to people who would otherwise not pay close attention to things because piloting an auto is frightening at first so makes a person receptive to whatever makes them safer and gives them confidence.

I've heard Defensive Driving described as driving with the assumption that someone else is going to do something wrong, and to be in a position to defend yourself from that eventuality.

The opposite is to drive Offensively (ha ha) which I guess is to assume that every other car will do the right thing and it's okay if I take a few risks.

I wouldn't say that at all since you're not assuming people are going to do something wrong per-se, you're more willing to accept that a situation that looks normal is going to turn out differently and be aware of how you can avoid any driver who's going to do that stupid thing.

I drive with the assumption that everyone else is drunk. Most friday night around here, it is true.

In the morning, I drive with the assumption that everyone is half blind and too aged to drive. I work next to an eye clinic, so this is also true.

Drive like everyone else is a fool and you should make it back home safe.

I learnt these techniques from the IAM [0] in the UK. Their manual is Roadcraft - the Police Driver's Handbook [1] which introduces techniques which I've found incredibly valuable.

[0]: http://www.iam.org.uk/ [1]: http://www.amazon.com/Roadcraft-Police-Drivers-Handbook-Foun...

Even beyond that, there's an overall 'field' of cars for me. Peripheral vision gets an overall sense of all cars in the proverbial neighborhood, and looks for scary patterns. Little things that seem harmless (and maybe are) put me on edge and make me guard my brake.

My wife didn't understand at first, but after 12 years of driving with zero accidents, it becomes apparent that it works.

...look at every car that I can see, using every gap, window, mirror etc to be aware of my surroundings. Look at people, bus stops, crossings, shops in towns etc.

In my experience a shortcut to this level of awareness is to regularly cycle in commuting traffic. Not only does one notice much more, but one also notices the classes of things that it's actually impossible to notice from an autocage. (Examples: what most drivers are doing in addition to driving, where pedestrians are hiding, how a line of traffic moves through signals in fits and starts, etc.)

I cycled 150 miles for charity once, you soon pick up on extra things - having been knocked before.

I came to make a similar comment regarding my hobby; skydiving.

I've been falling out of planes for over 20 years, and have seen the equipment progress in design and function such that people don't die due to the same things which used to claim skydivers. However the yearly rate of death remains about the same. The dark joke is that as the sport has gotten safer people have had to come up with new, more stupid ways to die.

Smarter people than me point to the theory of risk homeostasis. Basically, you accept a level of risk of an activity, if the safety of this act increases you'll find new ways to make it as risky as it was previously.

https://www.google.com/search?q=risk+homeostasis

In skydiving this is jumping smaller and smaller canopies (aka parachutes), swooping, base jumping, big way formations, aggressive angle flying, etc.

Not as dramatic as skydiving but when paragliding we had the saying: "There are old pilots and then there are brave pilots." ;)

The original form I heard that one:

There are old pilots and there are bold pilots. There are no old, bold pilots.

Yup that's it. I was remembering AND translating ;) Thanks

What about the overall risk on a motorcycle? I have two friends who were killed on motorcycles. Both were experienced, in neither case was it their fault, and I'm not sure if they could have avoided it. Seems unlikely. But what I do know is that if they were driving their cars, they would be alive today. I find people often correctly evaluate the risk of failure (low), but completely ignore the potential cost (your life) . I'm guessing they also have a different model for calculating reward, but it seems mostly an error in calculating risk and cost. BTW, these were IT guys in their 30s, not teenagers.

Certainly riding a motorcycle is risky. Lots of things people do for fun have very large downsides (death or otherwise). Some of us feel the risk is worth the reward. I totally understand if you don't agree.

In terms of experience - I see a LOT of experienced motorcycle riders that have "1 year of experience 10 times". I'm not saying that I'm a better rider than other experienced riders, however motorcycle riding is something that requires constant discipline (moreso than other forms of transportation that people generally do).

edit: Please don't take this comment as any indication of your friends experience. I'm speaking in generalities.

Just how dangerous do motorcycles have to be before the fun isn't worth the risk? Twice as dangerous as cars? 5 times? Maybe 10x? Surely not 20 times more dangerous?

From Wikipedia's article on Motorcycle Safety: Per vehicle mile traveled, motorcyclists' risk of a fatal crash is 35 times greater than a passenger car.

Motorcycles are insanely dangerous. There are 23 fatalities per 100 million miles ridden in the US. If you ride 4,000 miles a year for 40 years, that gives you just over a 4% chance of dying. Compare to a 0.1% chance for doing the same in a car. And that's just the risk of death. It doesn't include risk of paralysis, brain damage, or losing a limb. Even if you are a perfect rider, half of all fatal motorcycle accidents are collisions with cars. Many of those simply cannot be avoided.

People would rightly call you insane for driving a car that had 35x more fatalities than average. I fail to see how removing two wheels makes it OK.

(Note: I've been riding for seven years and had no accidents. There are some good reasons for riding a motorcycle, but they don't involve recreation.)

You are correct of course. I've ridden for 20+ years with only one fall in an empty intersection (gravel) and rode away with literally only a scratch. I'm pretty sure those days are over though as my wife is not a fan, and I recently sold my motorcycle.

Horse riding is about 10x more dangerous than motorcycles, according to stats I've seen.

Motorcycle touring is my primary leisure activity. It's what I do during summer holidays. A 300cc scooter is my primary means of travel in London. Riding on motorized two wheels is so integrated into my life that it would be hard for me to imagine a life without it.

It's almost a physical need. After a few days of snow, or a spending some time somewhere without a motorcycle, I start getting cravings. I'll frequently go out for a wander on weekends for no other reason than to ride.

Riding has a meditative quality. You can't think about anything else; you're forced to be in the moment, in flow, in a constant planning and reactive mode. Worries and concerns and any other busy thoughts disappear.

I've had an accident on a motorbike which ended up thankfully only with a broken leg and two months of immobility (the fracture was a bit complicated). It was not my fault and if I was in a car, I probably wouldn't have been hurt at all. Roughly a year after the accident, I started to ride the motorbike again, albeit not that often. I don't see anything I could easily improve with the driving style to reduce the risk. I think I'm already quite defensive both on the motorcycle and in the car and I try to always remind myself that I'm no expert.

Maybe I don't correctly evaluate the risk or I just assign too much value to the benefits I get. It's very hard to calculate with a cost of own life. Moreover this cost is changing. It can temporarily increase for example when you support a family. But how do you account for the fact that you have a finite amount of time here no matter what you do? There are much less risky things to do and also much riskier ones than riding a motorbike. Are all the people who take more than the smallest possible amount of risk crazy or unable to assess the values? Honestly, I still don't know.

Motorcycles aren't called "donorcycles" for nothing. The risk level seems to be intrinsically high, though it seems to be held up by bikers not wearing helmets:

http://articles.sun-sentinel.com/2012-06-10/health/fl-motorc...

The documented risk of dying from a serious head wound while riding without a helmet is so high, bare-headed motorcyclists are considered ideal organ donors: They're typically young, otherwise healthy and at high risk of brain death from sudden trauma.

I've noticed that some curve in software engineering and systems administration, at least from people who are trying to be serious.

At first, everything is new and scary, so they'll ask for help with even basic things, and maybe keep meticulous notes or comments about what they're doing.

Later, after they've gotten more and more comfortable, they'll stop keeping good notes and asking for help.

Finally, after being bitten in the ass a few times, they'll start to regard all computer systems as being in active conspiracy to make their life and their code miserable, and at last they'll finally have the correct mindset to get quality work done.

When I got my motorcycle license I was told that the probability of an accident happening increases a lot after 6 months / 3000 km (~2000 miles) due to the recently gained confidence. I have no idea if those stats are even remotely correct but it made sense at the time.

Also true for private pilots. People between 40 and 100 hours are pretty good, but from 100+ there is a rise in overconfidence.

More people need to read the "I learned about flying from that" series of books, which is a collection of columns.

I think the idea applies to cars as well, fwii the worst year for car accidents is the 2nd year after getting a licence (1st year being the worst), at least in some countries (can't remember where this is from)

Reply on news.ycombinator.com