1x54: The Trolley Problem

Jeremy Garcia, Jono Bacon, Bryan Lunduke, and Stuart Langridge bring you Bad Voltage, in which we are curmudgeonly, we are ethical philosophers, and:

  • 00:02:00 Are developers learning libraries and not learning the actual programmnig languages they've chosen? And is this a problem? Are JavaScript hackers just using jQuery or PHP hackers just using Laravel when they shouldn't? Or is this greybeards complaining about the kids today?
  • 00:16:47 Review: the Blue Yeti USB microphone. Almost by coincidence, the whole Bad Voltage team have purchased the Yeti USB mic from Blue Microphones, and so we all review it together
  • 00:27:20 The rise of self-driving cars brings up the question of algorithmic morality; how should the car be programmed in the event of an unavoidable accident? Protect the driver at all costs; reduce loss of life overall even if the owner gets the short end of that stick; what? This is a big decision that needs to be made: how do we think this should be handled?
  • 00:39:50 THe UK government have recently started making more noises about banning encryption from being used by ordinary people, to prevent terrorists from being able to communicate without security services reading it. It's the Crypto Wars and the Clipper chip, all over again. Meanwhile, Apple have made a big point of how they work hard to protect their customers' privacy by ensuring that iMessages are end-to-end encrypted and so forth. Clearly, these proposals are in opposition. The question is this: if Apple declared that these government proposals were incompatible with their customers' privacy and so threatened to pull out of the UK market... who would blink first? And would Apple do this? And is it OK that they might have this level of power?

Download the show now!

This may well be relevant to the self driving car debate when making decisions about human life and the complexities of programming morality: https://www.youtube.com/watch?v=7PKx3kS7f4A

In my opinion once matured the driving is likely to be in such a way that an unavoidable accident can’t happen without a third party causing it. In the cliff vs. humans scenario you had the solution is simple, drive slow enough so you can see upcoming humans and stop appropriately, that is likely the decision such technology can take and should be what humans are doing (yes I’m a kill-joy and don’t own a BMW).

I think there is a very typical tradoff when deciding wether to program close to the hardware or not. Time vs Quality of writing vs closer to metal. You basically can choose two, unless you’re some sort of code genious. If you want something done quickly, it tends to be easier to write good code in some sort of framework that does a lot of stuff for you. If time is no issue than yes it is better to write a good piece of code on a lower level, but if you don’t have the time taking it closer to the metal will have a bigger chance of producing bad code which might be worse for performance than doing it properly with some higher level language. So it ends up being a balance in which you have to choose what ever is most optimal for your situation.

However You can always take it a step too far where said language, framework, etc. Doesn’t actually make it easier/faster to program to same thing as it would be in a more close to metal situation. It that case ususually the issue is that people are too lazy to switch and figure it out a new thing, which in itself is a bit of a shame couse learning new stuff will make life better for everyone.
I’m not anything in the way of a proper programmer, so I could be wrong.

On the self driving cars thing. I think it will in the end be decided by insurance companies. Whatever solution is the cheapest will probably prevail, if it isn’t picked up by some mayor governments. Simply because they will create a monetary insentive for the system that saves them most money.

The biggest question in the issue about encryption is whether the politicians are really that stupid or just heavily influenced (cough… Corruption? Cough) by people in the survalance industry. I am not sure which is worse, but whatever it is people need to know couse they might do more damage than they’d ever foreseen…

I liked this weeks show again keep it up guys :wink:

You all agreed that no car company would accept liability - but that’s exactly what Volvo did back in October: http://www.autoguide.com/auto-news/2015/10/volvo-google-and-mercedes-to-accept-responsibility-in-self-driving-car-collisions.html

They did indeed. I stand corrected. On the other hand, I wonder if that’ll change the first time a self-driving car causes a death and they get sued for a hundred million quid?

I’m surprised @sil didn’t insist everyone use the Vanilla JS framework!

1 Like

The key element there is “if their self-driving vehicles cause a collision.”

And they are going to make damn sure they have more than enough data to prove it was the other person’s fault.

So sorry @jonobacon, directive number 4 will be “cover Google’s ass at all costs.”

I was actually going to mention it as a nice example of appealing to people who want a framework but actually shouldn’t :smile:

In John 15:13, Jesus said: “Greater love has no one than this: to lay down one’s life for one’s friends.”

So I’d say Christians who follow Jesus should want the car that saves the pedestrians rather than themselves. Not everyone wants the “save me!” option :smile:

I wonder, though, whether there’s a social solution to all this. Simply require a clear marking on the outside of a car, perhaps on the license plate, as to whether it has a “save me” or “save others” policy. It would be very interesting to observe the social dynamics of that.

Agreed not everyone wants it. There’s massive social pressure to ask for “save others”, to be perceived as altruistic, but equally massive personal desire to ask for “save me” (for most people, I would suggest) if one can do so secretly. Your point about the social dynamics is a jolly interesting one; one way that the environmentalist impulse started to take off in cars is when it became perceived as cool to have a Prius. If “save others” (or, more accurately, “don’t prioritise ‘save me’ over everything”) can be given the same patina of coolness, then maybe this might just take off (and then it wouldn’t just be those following John who would choose it!).

Regarding cars saving pedestrians, (possibly killing the driver in the extreme case).
This only has to have a few false positives, where a pedestrian is incorrectly identified and the driver killed - for the idea to become unpopular AFAICS.

Just wondering how well computer-vision could differentiate a kangaroo standing upright (on a rainy/foggy night for example), from a pedestrian (this isn’t a contrived example, kangaroos get hit by cars all the time in Australia).


Another factor is how reckless the pedestrian is being:
If someone wanders out onto a highway today - they risk their own life (mostly). … would a scenario where reckless pedestrians (mostly) risk the drivers life be better?
On the other hand - these are computers making decisions - so they could easily plough into a group of people if the “reckless pedestrian” condition was met.
(Of course it would try avoid fatalities in both cases…).

I recently returned from OggCamp where secrecy and encryption was discussed in a number of sessions including one where our friend Mr Langridge was on the panel. To be fair the proposed legislation does not seek to ban all encryption only strong encryption: meaning that you can have end to end encryption provided it is done so that the authorities, and by extension anybody else who can intercept your traffic, can decipher the message.

This is of course plainly stupid and will not stop organised terrorist or criminal gangs who will just continue to use encryption illegally: If you are prepared to commit murder for you screwed up beliefs then I can’t see that making encryption against the rules stopping anyone. I believe that privacy is very important and implementing secure communications is not a technical problem but a social one. We need to convince people of the importance of privacy and convince people to demand it recognising that many of the services we get for free today are paid for by mining and selling our data so for example companies don’t waste valuable time and money trying to sell me a product I am never going to be interested but target that time on people who they may make a sale from. This means we are going to have to start paying for these services.

The problem here is a social problem getting people to care about their privacy. This is not easy and perhaps we need to show that encryption is a tool that can be used for good as well as evil and pointing out a few examples.

Without encryption for example there can be no online shopping as everybody’s credit card details would be known by millions.

I can think of hundreds of others here and I wont list them as I am sure every one else using this forum can too.

The UK government is trying to take the line that bad people are using encryption to do bad things and we must stop it. This argument is throwing the baby out with the bath water: Some bad people use a carving knife to commit murder but I don’t see any calls to ban me from cooking a Sunday roast.

Finally @sil thanks for your time to make me feel welcome at OggCamp, much appreciated.

Excellent episode again guys, a joy to listen to. Not to get competitive, but the thoughtfulness and research that clearly goes into these episodes makes them so enjoyable compared to other offerings in the podcast-sphere.

Few things I think are worth mentioning.

On the topic of frameworks, I thought this was an interesting section. Programming/Development sections haven’t been a huge topic on bad voltage and make an interesting change. When it comes to frameworks, seems like there is a few things to consider:

  • Tradeoff between performance and complexity reduction. If a framework helps me do something which is complex or requires a lot of boilerplate with only a small performance hit, then I think its the correct decision. The more performance you want, inevitably the more layers of abstraction you must peel away (right down to things like loop unrolling etc).
  • use of frameworks in general are almost universally revered as a good thing, however overuse or sole use of a framework is obviously a bad thing, as it leads to square peg in a round hole style development.

On the encryption point, I just don’t get this. It makes no sense… Encryption is such a fundamental part of every company in some way or another, not to mention the internet as a whole. I don’t get why someone hasn’t shut this crazy parade down a long time ago. The idea that this would be anything more than a proverbial fly in the windscreen of oncoming terrorists is laughable. Unless we are underestimating the stupidity of the people terrorizing our lands.

Finally, Jeremy mentioned that Jono has a new interest in behavioural economics, (unless my ears deceived me), Is this a real thing? I’ve recently become quite interested in this topic, and just wondering if you have any reading/watching material you would recommend on the subject?

I’d recommend everything by Rory Sutherland. I believe Jono has some book recommendations too.

1 Like

“Freakonomics”, “Predictably Irrational”, “Outliers” and a few Clayton Christensen books/essays are what initially got me interested in the topic. “The Paradox of Choice”, “Nudge”, “SuperFreakonomics”, “Influence: The Psychology of Persuasion”, “The Upside of Irrationality”, and “Thinking, Fast and Slow” were all worth reading as well, IMHO. “The Black Swan” leans more toward the economics side but is also a solid read. I’m sure I’m missing a few, but off the top of my head that’s where I’d start.

–jeremy

1 Like

Comprehensive list… Thanks very much!!

Morning all, I was looking through Amazon UK’s “Lightning Deals” this morning, and saw that they have a deal on the Blue Yeti Microphone with a whopping 40% off. That makes it £69.99 with free delivery.
The link is: http://www.amazon.co.uk/gp/product/B002VA464S?ref_=gbps_img_s-3_8627_8e40b4aa&smid=A3P5ROKL5A1OLE

This deal ends in 4 hours, which makes it 13:00 on Wednesday 18th November 2015.

So… Little 4 year old Johnny is pus for using Scratch, MAN UP AND USE ASSEMBLER YOU RETARD!

Sheer luxury real coders work directly in hex :smile:

Only joking, there are times when you need to get close to the hardware or the maths is sufficiently critical that you need to optimise your code to within an inch of its life but there is also nothing wrong with using frameworks and high level languages if it gets the job done.

Please respect our code of conduct which is simple: don't be a dick.