Tuesday, 30 October 2012

Better than an average

Better than an average


20 years ago we were talking about how mobile phones were really going to revolutionise our lives. They were the latest thing. Today they are still the latest thing but now they are becoming a more seamless part of our lives. They are absolutely transforming the potential possibilities available to users, but note the word "potential".

Our local council has just polled residents on the subject of what the speed limit in certain streets should "be".
A good PR move for the council as it can be used to demonstrate the council is listening to and appreciating the public and confirming that they can have an input into what will inevitably become legislation .
In theory it will protect residents from high speed motorists. From a Tech. perspective I think dynamic (variable according to weather, time, location etc) speed limits are a great idea. 

The poll achieved about 400 responses, something the local newspaper jumped on and immediately conducted their own poll from their news website, which naturally attracted more participants and not unsurprisingly, quite different results.
Not unsurprising because there will be a difference in the motives people have for participating in the survey on the news website vs the council website.
It's not so much a reflection of the demographics of the websites. I think it's about the  motivation for completing the survey.
In some ways it wouldn't matter what the council did, thousands will vehemently oppose it, which kills any chance of intelligent public ideation.

But is this mystical "appropriate speed limit" based upon on who is perceiving it? Of course it is. Should it be ? I'm going to say no.

If you were to ask people with kids, as opposed to those without, what a speed limit should be on the road they live on there will naturally be a difference in opinion. The minute a family has children, cars appear to travel at twice the speed. You could also rightly expect locals would typically want lower speed limits on the roads around them. These are obvious factors but appreciate that in any survey the respondents decisions will be influenced by literally millions of factors, in this case it might be age, sex, driving experience, if they have had accidents, seen accidents, those who drive vs don't and, that's my point.
There are so many individual factors, and behind each factor a perception , there will also be a lack of empathy for those with different perceptions and hence different opinions.

Sound familiar: religion, money, war, politics, anyway one problem at a time.

I can't help thinking that we could use technology to help sort through all these social and environmental factors and ultimately build an "empathy logarithm" that could teach empathy for various situations.
If your oxymoron sensor has just gone off its because computers don't have empathy so how could they possibly study it.
Well, I believe we could we use technology to teach people "how" different people have different perceptions to identical environments and some reasons as to why.

Machine learning and problem solving could be used to help apply some logic over the issues, it could be used to help people consider other peoples perceptions in political social and business environments and would be one of the fundamental improvements technology could deliver to our lifestyles.

But first...

Since when has the general population ( and no, I'm not trying to offend the entire driving population ) actually had a knack for knowing the appropriate speed to travel down a road ?
You could argue that 99.99% of the time drivers don't have accidents, so they must be fairly ok at it. However, I'm going to suggest that all of that changes when the proverbial hits the fan.
Having sat next to about 2,000 people just before they each had a small, managed, low speed car accident. (Yes - I'm a really bad passenger, but all were in a training environment) I can tell you that the vast majority of those people had no idea at all about vehicle dynamics, that's the basics of weight transfer around the suspension, the weight on the tyres during braking, turning and accelerating, how long it takes to stop a car, what "their" car weighs, the size of the tyres. etc.
What's even worse is there is no requirement for a driver to know if their vehicle is front rear or four wheel drive, has ABS, traction or stability control and to be honest, the plethora of three letter anagrams that no one has ever tried to standardise or simplify, make it even more confusing for a motorist to understand what functionality to expect from such systems in different brands of cars. So how can people know what to expect, when it all goes wrong, well, they can't.
Hence, there are a very small handful of "Advanced driving schools" usually run by retired Motorsport competitors, that attempt to fill the gap, teaching the basics of car control and the preventative skills of using long distance vision and vehicle placement.
If you haven't done such a course, do one. If you have, you'll probably remember thinking it a bit weird that such courses were not mandatory.
Road safety experts, usually statistically versed, typically, have little regard for such training, sighting up-skilling as a danger rather than a benefit.
Monash University in Melbourne, where the vast majority of our road safety thinking emanates from, has made repeated attempts to furnish statistics to prove that up skilling causes over confidence. A rather broad but statistically possible generalisation.
But the whole situation is odd ,because education and preparedness seems to be a pretty important factors in every single facet of life, especially dangerous ones, well, except for driving ?   

As a system, we gave everyday nice people a licence to hurtle a 1,500 kg piece of steel around in almost any direction they choose, at up to 30 meters per second, without requiring them to have any education or psychological training, other than, the road rules and some very basic movement skills like turning and parking. In addition we also have humans doing the same thing but from the opposite direction with potential impact speeds of 200km/h with a 15cm wide strip of delineating paint separating them (Just imagine OSH giving the green tick with those same figures, human controlled, in a factory. It would never happen.)
Yet at the same time as failing to provide them with a relevant education, we told them certain brands were "safe", that "road" and "safety" could belong together without being quintessential oxymorons. 
Next time someone tells you to drive carefully, well meant I'm sure, ask them "How?"
The reply will be something along the lines of "keep within the speed limit, wear your seatbelt , don't drink and drive."
People assume, because of the messages we've been sending them, that they are safe as long as they are travelling within the speed limit.
Even car crash testing is made to look "safe" the vast majority of the car crash tests you see on TV are conducted at a maximum of 64km/h, that's all that the major crash testing standards require.
No Car maker in their right mind would show you their cars being crash tested at 100km/h.

Now, with the assistance of the internet, a dose of reality.
Here is a two car head on collision - each car traveling at 100km/h (62mph)  both cars were driven by remote control, no lives were lost.
Volvo and a BMW, a bit old now but statistically safe and sturdy cars

Here's a slightly more modern car, still remotely driven, into a crash wall at 192km/h: e.g. same impact force as a two car head on , each vehicle at 96km/h


Welcome back, Is 100km/h a safe speed to travel at ?

If the speed limit was there purely for the purposes of ensuring safe impact speeds it would be about 25km/h. So forget 100km/h as a safe speed.

Whilst this crazy situation is not the fault of any one individual, although one would think intelligent humans should be more curious when it comes to understanding something so dangerous, that they engage in on a daily basis.

Average opinion

I'm not saying the general public don't know what's best for them, and let's make it quite clear for the record, I'm a big fan or democracy - but even I would argue that it's probably better for a road safety expert to make such decisions than an uninformed public. Otherwise we could just save a whole lot of time and count the number of kids and elderly that live in each street from census data and set speed limits according to that.

So with all of the psychological and physical factors that are facets of driving, is the best way to set a speed limit by asking the public for their opinion on what a speed limit should be.
We are asking a collection of untrained people to make decisions that even an expert would fail to quantify.

Ask any motor sport competitor what a speed limit should be on a certain road and you'll probably get a dumb look, that's because they don't really think like that.

The motor sport competitor is constantly aware of and looking for all sorts of factors that have to be constantly kept in check to get them around the track, whilst motorsport competitors do this largely for fun, it's a tough job and for the drivers, they are constantly checking their own psychology to on one hand keep the car smooth, balanced and pointing in the right direction. They are quite critical of their own driving. It's actually not rocket science but it gets 100% concentration 100% of the time. Motorsport competitors know that a large moving weight, is really dangerous, some of them have lost fellow competitors in the sport, they see accidents on a more regular basis than road users and they tend to be more mechanically minded so they have each done quite a bit of thinking about the physics of what they are doing.
Of course, they wear helmets, 5 point seatbelts and before they start racing have quite a few years of club level experience. Still, despite all the safety gear, the lack of traffic, the huge tyres, the roll cage, the better suspension, their elevated skill level and practiced thinking, they concentrate hard. Really hard.

These people however will be the first to tell you there is no magic safe speed for any road, its variable. It changes, all the time. The drivers job is to monitor the environment and make decisions through visual observation, feeling, thinking and finally by control, quite a complex task and it would not be made any easier by diverting the drivers attention to an instrument on the dashboard every 5 seconds as prescribed and advised by current learn to drive practitioners.

Let me also just say at this point I don't think the use of excess speed is a good idea on a public road, at all. I don't think the speed limit should be higher. We all need to drive at roughly the same speed to ensure flow. That speed should give sensible margin for unexpected errors. But before we bath  in the ideals we should consider how it's possible for a driving population that didn't get the right training, to judge the right speed to travel at.

One comment I heard a lot from motorsport people was that drivers on the road travel way too close together, too close to the centreline and too fast in the wet/dark or in poor visibility.
Yet the average motorist would be blissfully unaware that the slower car on the road might be driven by a competitor who has assessed the situation much differently.

Speed is an easily measurable factor , hence it's easy and dare I say it profitable to legislate against certain levels of it.
Concentration levels which ultimately dictate the actions of a driver however, are not thought of as measurable.
Pity - because most drivers are psychologically at risk of failing to concentrate. You might ask how this can possibly be when the physics of driving suggest the activity is really dangerous.
The drivers concentration isn't as intense as it should be because there are less regular reminders of the speeds and forces present and no natural signs of risk, and, it's your fault, bear with me: read on.

Motorsport competitors seldom, "forget to look", or "doze off" or "lose concentration" because they are highly aware of the environment. They don't fiddle with the stereo or send texts because that would be well, screaming dumb .

Tightrope walkers also, seldom lose concentration, they have a natural reminder, all the time, height
In a modern car however the danger is obscured especially to those who are unaware or disinterested in driving.  

Power steering, modern suspension , radial tyres and years of vehicle design and noise insulation mean a modern car is now a quiet cocoon of luxury rather than a place to recognise and evaluate deadly risks. 

A quick look at our roads gives you another clue. Long straight roads are in essence boring, they don't require the driver to do much, so they don't, drivers become accustomed to having less to react to, they switch off , stare at the vehicle in front and become accustomed to just hanging on to the wheel. Fatigue wasn't as much of an issue when drivers had to manhandle older cars around  twisty roads.
Today the second anything goes wrong we blame the roads, we spend millions re-engineering perfectly traversable roads because drivers forgot to concentrate, they failed to identify the risks.

You might be relieved to know you only have a small chance of having a serious car accident in your life time, this is another reason why people are not as alert as they should be. However statistical measures are of little consolation to those interred or living highly compromised lives as a result of crashes.  How many crashes ?
http://www.mixnmashentry.blogspot.co.nz/

Dynamic speed limits

Change with time or location, or any other measurable factor so perhaps for every road the  speed limit at 3.00pm or 5.00pm  should be different to the speed limit at 2am in the morning. A classic example of this is school zones.

If you are old enough, you might remember the LSZ's Limited speed zones, back in the days where everyone had to think for them selves you could assume the speed limit within these zones was dynamic, it changed between 50 and 100km/h according if the road was wet, the visibility was poor or whatever, the speed limit changed. The warning sign did not change, just the speed limit based on your own perception. LSZ's are gone now. replaced by blanket speed legislation.

Commentary Driving

This is a great exercise for drivers, It now makes up a small component of our licence test. It requires a driver to verbalise all the factors they can see/hear and feel on the road. It's a great way to test a drivers awareness, evaluate how far ahead they look and their overall perception of the risks. Here's a good example of a young driver performing a commentary drive who you might notice is much more aware of his surroundings than what we might expect to see here in NZ.

The App Idea-

We're getting to the point where significant percentages of drivers carry gps navigation apps on their dashboard, you know the duplex ones that can transmit data as well as receive it. So considering the opinion poll achieved less than 500 responses and then essentially “averaged” the results - would this high tech confusery we call a smart phone have been able to offer a better solution.
I would gesture that in a matter of days, it would have been possible to make an integrated  solution with a Navigation app, allowing users to vote on the speed limit restriction for the road they are traveling on, and of course say WHY, with respect to time, weather, visibility, proximity to the nearest school etc etc.
At this point you might be thinking. To input all those factors is going to be a nightmare for a user , it will just distract them , make driving dangerous etc, but the user only needs to input one thing. VOICE.
an example:
A driver travels down X Rd at 6am presses the screen, or perhaps some sort of VOX activation and states "thirty traffic density sun strike".

So even at the most simplistic build level an app could record the time and location from the gps sensor, direction of travel from 2-4 consecutive readings and a voice note like an mp3 or wav, to a human (that's a job)
Voice recognition is getting really good now but the paid human can just take the voice recording from the server and extract the users "reasoning" for their proposed limit.
The system could then analyse things like the time, which gives you the day, position of sun, the season ,school terms, historic traffic density, traffic cams, etcetera. Add weather, wind direction and speed, local events and any other data that could become a factor used to help qualify the user suggested speed limit. Just for the record it does not have to identify who the driver is or how fast they travel.

Instead of our poll just collecting a large number of "suggestions" for a speed limit we now have lots of "reasons" for those individual speed suggestions. Those reasons (essentially factors) will change according to public perception, time, weather etc. So after perhaps 100's of speed suggestions from users, a profile could be developed for that location/road, with all the factors taken into account. Some factors will only appear at certain times, eg sunstrike/ kids on bikes/event traffic. Some will occur according to the weather, It's not a huge stretch of thought to do this for every road in the country. Would analysis of all these factors give us a better understanding of the dynamic roading environment and how it changes, naturally it would.

I'm not even going to touch on the massive advantages of feeding this data straight back to motorists but readers might see some logic in a post I made some months earlier when we made the giveway app.



So instead of using all these factorial observations and looking at when different factors emerge - should we just ignore the factors, take the average of the values of the speed suggestions and let the status quo become the new speed limit, of course not, but, that is what we do sometimes when we crowd source answers to questions , well meaning pre defined questions, and then "average" the values of the answers out.
In a way that's how voting works, we get an average of opinion, but that "average" ignores ALL the ideation and factors that went into each answer.
Ideation is not something that benefits from a mathematical average.

Instead should modern polls collect the variables, study their relationship to each other and let their dynamics of the data speak for its self ? 
The internet is probably guilty of ranking user opinion in such a logical way (1-5 ratings, likes and dislikes) which is easy for computers to analyze but I wonder perhaps if it potentially trains people to have easily polarised thoughts. Only on forums does ideology actually emanate, the TED forum is a great example of intelligent people who appear to understand empathy and ideology. Youtube, depending on the topic : not so much.

So could advanced Algorithms and machine learning take all these factors, and turn driving into a commentary competition. Where drivers could get points for noting existing and new factors about the road, as they drive.?
Rather than just using the factors to warn other drivers, could we make the driver work to identify hazards and reward those drivers who are being more observant.
If we could do that, that really would be a measure of just how much the driver is concentrating on the environment.
That's never really been measurable before.
What if we could learn that vehicle positioning, following distance and concentration levels are far more important factors than using up the drivers visual resources by shifting their focus from the road to the dashboard every 5 seconds in an attempt to adhere to some "magically safe speed limit"?

You would hope that since about 5,000 people seriously mangle themselves in cars each year and another 300-500 lose their lives, costing the country in excess of 2 BILLION dollars each year, that by now, there would be a whole flock of road safety experts looking into new technologies that can help this sort of thing,  but no. Instead were putting money into re-designing perfectly good roads, and into revenue gathering devices that make very little difference to what people concentrate on.

We're a smart country. I'm sure I'll get a phone call from an oxymoron expert soon and when I do I'll be very happy to provide ideation at no cost to such a good cause.

Driving IS dangerous, and there is no such animal as Road Safety.


James Hancock
i.deas.digital.consultants