I think this is important to remember in our field too. Too often I hear about a user story breaking down because "the user wasn't doing it right".. blaming the user is the wrong way to go -- you just set it up wrong.
Sometimes the user is wrong. If a person drives their car into a lake and drowns, do we say that cars are broken because they allow this? Or do we point to the millions of other drivers out there and say that driving on roads (and not into lakes) is the expected user behaviour?
An extreme example, sure, but I think it also translates to software. Some people criticize vi/vim for being modal because it confuses beginners. They use this as the basis of an argument that vi is a bad way to edit text or that modal interfaces are bad in general. But tons of other people take the time to learn the vi system of modal editing and they love it so much that they write plugins for everything else in an attempt to replicate that experience. Who is wrong here?
You can build powerful systems without confusing people who often never intended to wind up in your program in the first place. You can even build them to show people how to use them.
Cliffs, lakes, and highways often have guardrails, even if millions of cars drive by without going over the lines.
Vim, IMO, is different because it's not meant for consumption by the general public. Vim is a sysadmin tool, in unix-like environments, where the users have a general amount of competency before even encountering it. The whole rat-experiment analogy is more about stories that are public facing.
Users will always find a way to do what they need to do. The challenge is in finding out what they need to do because most of them would rather come up with their own solutions than come to the developers with a fully-specified feature request. I think this is why spreadsheets are so popular. They are never the best tool at any job, but users can bend them a million ways to do what they need to do.
People race trains all the time because they don't want to sit for 120 seconds. The problem is that losing that race means death.
The is especially stupid for the trains cited. "Metrolink" is not some slow cargo train that you will be trapped behind for 15 minutes. A "Metrolink" is about 6 train cars and is generally moving at a decent clip--probably no less than 35mph. You're not going to be waiting long.
LA has a much more significant problem with people walking in front of or committing suicide on the train tracks. Those cases generally far outnumber traffic accidents with the trains.
That's funny. The very first thing you learn on a ship is to watch out for exactly this. Any ship that maintains it's position relative to yours is on a collision course!
I like the Dutch approach here: replace them entirely.
https://www.bd.nl/oss/vier-kinderen-4-tot-11-jaar-omgekomen-...
In my native language, they still haven't really found out a satisfactory way to differentiate those.
I don’t know why grade crossings without gates do not also have stop signs. We have to assume that many drivers are unaware of the rules of the road, either through lack of training, forgetfulness, local culture, or inconsistencies with other countries.
But as this article points out, you can't use your experience judging car distance/speed to judge train distance/speed.
From a fairly recent HN article -- see Lamport's discussion of Buridan's principle.
Now, take a 50 meter long airplane or 100 meter long train. Or a 300 m long ship. It's not going to work - fatal accidents ensue.
//edit: okay, the above is a bit minialistic, so, in full: The article is a good point on how systems have to adapt to actual human (mis-)behaviour, instead of relying on wishful thinking on how humans should behave.
I bet it has something in common about hand washing in hospitals taking some hundred years to become mandatory - or how it was done much earlier in Hungary, yet nobody outside believed in it.
https://motorbikewriter.com/smidsy-biggest-cause-crashes/
https://motorbikewriter.com/scientific-studies-explain-smids...
Most drivers find it very hard to estimate the closing speed of a 2 wheeled vehicle approaching an intersection.
Cars have become a lot safer in that time. And the number of crossings changed by a factor of 0.68. Both of these could account for the reduction in railroad crossing deaths.
I would guess that roadway signage in the US runs about 80:20 textual:symbolic, but that the EU runs about 20:80, with much heavier use of symbols, and use of text only to modify or explain something not covered by the available symbolic palette. Could any EU->US or US->EU migratory drivers make their own estimates?
They're only victims of their own stupidity. The real victims are the train drivers who are often traumatized, injured, or even killed in these collisions through no fault of their own.
The same goes for other occupants of the car. Pretty much everyone but the driver.
Rule #1: if you are at a railway crossing and you see a large object in motion coming down the track, then stop and let that large object pass before you try to cross the tracks!
And why should they? Some people die, but how many times more people do cross successfully?
That seems to be a weird way of describing "steer so the target is always in center of the image".
There’s no need to estimate position and speed of objects and then extrapolate them to predict whether course collide and how long that will take. It’s easier to directly derive that from optical information (think of it this way: if that train is 10% as far away, and also goes 10% faster, it will still hit you at the same time. That’s a degree of freedom brains don’t have to solve for)
(Couldn’t find a really good link. You’ll have to do with https://en.m.wikipedia.org/wiki/Ecological_psychology, http://www.scholarpedia.org/article/Optic_flow#Active_and_ec..., https://www.latimes.com/archives/la-xpm-1995-04-28-mn-59922-...)
There are three different examples near me of crossings they haven't closed with very different scenarios.
1. A large industrial estate built on the tidal river is reached via a road only barely above sea level. The road is used by lots of heavy articulated vehicles (it is after all an industrial estate) yet it crosses a four track railway that serves both freight and passenger traffic to a major city. Result? A major at-grade crossing which is closed to road traffic for about 10-15 minutes at a time as much as three times in an hour. This crossing has full barriers, and a pedestrian bridge (so pedestrians needn't wait for the barriers). There's no way to dig down (it'd flood) and a road bridge would have to be very large to carry those articulated trucks but there's nowhere to put it. I expect this crossing will remain until forces of gentrification some day mean the industrial estate turns into housing and they just close the crossing altogether.
2. Where that same railway once went through the city to deliver rich people to the ocean liner terminal itself the passenger trains these days are diverted through a tunnel under the city - however a few freight trains per month still make the journey to the docks. So a wide city road goes over an at-grade crossing that is rarely needed, most users probably have no idea it's a crossing. This is a full gated crossing, but it's manually operated. A crew will come out, switch on stop lights, block the road, move the gates, then one train drives in or out of the docks, then they unwind everything. Because it's so seldom used a bridge seems unacceptably costly, and because it's manual I'd guess the residual risk is low so I expect this to remain in use essentially forever.
3. A rail branch towards another coastal city cuts across a residential street I used to live on. This one I can imagine closing. It's currently using full barriers but has no pedestrian route except to just walk to the nearest bridge. I think sooner or later somebody will get themselves killed, clambering over the barriers or whatever and they'll just close it because there's no way to make it any safer other than closing it.
Edit: I'm not trying to but any blame on her. It's easy to think in hindsight of what could have been done, another thing entirely to do it in seconds while panicking.
My point is that even with all these insights into how human perception works, I agree with the strategy that it's best to bypass railways completely by going under/over them. And if it's doable in a densely populated country like the Netherlands then it's doable anywhere. (Some of the underpasses that have been built in recent years have required significant changes to the whole surrounding layout of the town, like one they completed in Bilthoven near Utrecht. Yet it was done.)
Sigh.
Railway crossings are pure evil. There is no fail safe design that I know of without adding a proximity sensor inside the car (I am a railway signalling engineer). Would love to be wrong though.
Confirmation comes from observation by a signaller or crossing operator either directly or by CCTV, or automatically by LiDAR and radar. These crossings tend to be used in areas of heavy road, pedestrian or rail movements, and when train speeds may exceed (from memory) 100 mph / 160 km/h.
Their disadvantage is additional installation, maintenance and operational costs. They must also be closed for longer before a train arrives than an automatic half-barrier crossing to allow for the required safety checks and ensure that signals are cleared far enough along the line that approaching trains do not need to slow down.
All safety mechanisms are this. The tricky bit is the value of life calculus. Is it worth an added $10/year tax for every taxpayer to MAYBE save 1 life per 5 years? You'll get different responses for different reasons.
That $10 may be a rounding error for many, but to some its the difference of a week of lunches for their kid or not, and they vote, too.
The big issue with trains is that it has an inertia that our brain is not trained to.
As signalling engineers, we get to compute safe braking distances a lot and they are /big/, even at relatively low speed. This means that trains in curves -> do not have line of sight all the way to the next possible stop point <- so the driver may never have the chance to stop if you fall while crossing the track.
So don’t do it. Even at 30mph in curves, you are in danger.
"Human error" isn't a very useful "root cause" for any incident. Humans make errors.
But if there's a working automatic barrier, and the user bypasses it? Tough to see how you could UX your way around that.
https://www.fizzicseducation.com.au/articles/science-in-publ...
Is that clear? Most people don't get hit by trains, so I don't think that's clear. I'd say most people are cautious around trains while a minority aren't, and a minority of those who aren't will get hit by trains.
I also find it amusing that all of the stories regarding train-car interactions seem to occur at crossings with modern signaling devices, rather than "at your own discretion" crossings in the middle of nowhere. This leads me to suspect that the typical human has been conditioned to treat any signalized intersection transition as a "beat the yellow" event, but in the case of a rail crossing the brutal physics equations seem to be conveniently ignored.
Perhaps there needs to be a day in driver's education courses where everyone has to review just how heavy a freight train is and how much kinetic energy must be dissipated in order for it to come to a complete stop. Maybe make them ride at the front of a train to experience how painfully long an emergency braking maneuver takes to complete.
This kind of accident is so common I remember the government commissioning a series of funny PSA ads that played on national TV showing people acting recklessly at railway crossings, getting killed and then expressing their regrets from beyond the grave.
A large part of this is due to population density (or trip density) around the crossing. If you've got 1000x the number of trips on a crossing, a 100x safer crossing is still 10x more likely to result in a colission. Fill in your own estimates, as my numbers are clearly made up. Also, guidelines require more signalling in built up / dense areas, and at intersections with a history of colissions.
Additionally, if you have a colission at an unguarded intersection, the response is often to consider a guard, and if it happens again, the story is 'senseless officials refuse to put in guard' rather than 'sensless person disobeys guard and is injured/causes confusion and delay'
Personnally, I belong to the school of « remove them all, whatever the costs ». A little extreme, but €&@% I hate these things.
Edit: Clarified that it is written test.
Those that ignore the signals and live, that is.
There's three different "doables" at play here:
- technically doable: can you actually convert the level crossing to an underpass or a bridge? For example, it's hard to do an under/overpass when there is a river 10m away, or when everything bar the road/rail and a bit of space is filled with buildings.
- financially doable: is there actually enough budget in the city/responsible authority's coffers?
- politically doable: will the population support the necessary construction works and the limitations that come with it? Like, will there be heavy protests of people that rely on either the rail or the road to get to work and would require massive amounts of time to bypass?
It is rare that there is a combination of all three being true, with the exception of course being tragic disasters with people and especially children dying as such events usually eliminate the political and financial blockers.
They have disadvantages including expense, and additional road closure times (compared to automatic half-barriers) as they must be closed and proven clear early enough that an approaching train does not need to slow down.
ADDED: And while "out" is obvious with an exterior door, it may not be with an interior door (and in fact there may not be a clear "in" and "out".)
146 people died in club in France due to inwards doors in 1970: https://en.wikipedia.org/wiki/Club_Cinq-Sept_fire
I don't know the scale of the fine but using inwards doors are a big no-no.
And similarly, electrical door must unlock themselves when losing power (unless special buildings like military or dangerous laboratories).
If you instead steer so the target’s bearing doesn’t change, and ignore whether that bearing is in the center or somewhere else, the missiles are much more likely to hit.
Instead you want to intercept the target. You want to reach the point where the target will be, and at the same time as you. So keep the target to the side, at a constant angle.
There's a physiological reason why people have a tendency to underestimate risk in certain situations, or miss obvious pending collisions. The usual comment from someone who has had such a collision is "They just came out of nowhere". See also https://news.ycombinator.com/item?id=23024275
Airplane pilots (particularly military fast jet), and ship captains are taught specific techniques to deliberately and systematically overcome these physiological limitations. Some of the techniques are taught in some driving schools (for example https://news.ycombinator.com/item?id=18157090 ) and to be frank I think driving education should go all-out on it.
Might I still be hit by a train I never saw at all? Sure. Somebody I knew in highschool was hit by a train on a track he believed was abandoned and consequently, never looked down (he lived after being pushed 100m down the track). It's possible that might happen to me, at least if I ever forget that it happened to him. But what won't happen is getting hit by a train I see and judge to be moving slowly or not at all, because I stop for any train I see no matter how much time I think I might have to cross the track.
> Airplane pilots (particularly military fast jet), and ship captains are taught specific techniques to deliberately and systematically overcome these physiological limitations.
The "specific technique to overcome this physiological limitation" is: Don't race trains. Don't race trains even when you feel certain you can easily beat the train. Racing trains is stupid.
Nobody cares what you do all the time, though. In order to reduce accidents overall, we care about what the average person does. And if the average person does something else, then common sense isn't really the right term.
And even average person isn't the right bar. To really cut down on accidents, we need to see what the person who's behaviour is a couple of standard deviations less safe does in various scenarios.
export EDITOR=/my/fancy/editorIn Windows if you open a new filetype for the first time, it asks you what application you’d like to set as the default type, and on top of that it tells you where to go to change that later. That’s what beginner friendly UX looks like, not just throwing the newborn into the pool like Linux does with Vim editing.
If the "newborn" in 2020 insists on using the terminal instead of the much more beginner friendly options that are the default nowadays... granted, but why compare a terminal editor in unix to a GUI? That is like comparing apples and oranges and then complaining that the strawberries don't taste like bananas.
The things is, I don't remember the written test I took having any content at all about railroad crossings.
Physics tells us that we should accelerate to spend the least amount of time in the danger zone (rail bed). Since the velocities are orthogonal, it doesn't matter if you are hit by a train when your car is going 55 or 5 MPH (probably, unless trees in the resulting vector of your now pulped car).
It never made any sense to me to see school busses stopping at a grade crossing, then taking an extraordinarily long time to cross those tracks. What happens if the engine fails while on those tracks and a high speed train appears.
No, my vote is to make sure you have a clear view of the tracks in both directions and punch it. Less probability of interacting with a train.
As an example of how manually configuring the editor flag can be annoying, consider a job where you regularly ssh into many machines, each that just by default use vim. I know how to use vim, I just don't like it, but pretty much every single machine I have booted into always assumes I want to use vim, and even if you remember to change the flag doing it to many machines all the time is majorly annoying.
It's quite evidently you don't think the average person is the right bar, because the average person doesn't race trains. 3000 a year is next to nothing compared to other common hazards.
Maybe we should eliminate all intersections, since although the average person obeys signal lights and signs, the occasional idiot chooses to run red lights instead. We should eliminate crosswalks too; the average person looks both ways before stepping into the street, but occasionally people don't (this happens FAR more often than train/car collisions.) Let's ban bicycles too; the average cyclist wears a helmet but some choose not to and get their heads split open. Let's also ban self-service gas stations; the average person can pump their own gas without burning the station down but sometimes some people fuck up and cause huge fires. Doesn't that mean we should all follow the lead of Oregon and New Jersey and ban this activity? 'Nobody' cares that you can pump your own gas safely, please be considerate of the rare idiot who can't.