Traffic Tinder
How we all decided the rules that we do not follow
TLDR Want to watch videos of traffic offences and be the judge? Jump to the form at the end!
Here at Supervaisor we’re interested in traffic risk. In fact we’ve already collected over ten thousand videos of traffic offences in Tallinn with the help of our community. While we continue to grow our data capture network, we also want to test the potential of various forms of community based enforcement.
Here’s an example video we used as the first test to poll opinions.
For this particular case, 60% of the respondents thought this behavior should not be punished.
Posing the enforcement question as “Would you feel safe walking with a stroller on the walkway next to this behavior” dropped the support of this behavior to 43% yet still remaining surprisingly high.
Those of us in democratic countries have elected political leaders who in turn have passed laws that, among other things, govern our behavior in traffic. In essence we all took part in the process through which we agreed on the rules that we all should follow. Yet, when it comes to the enforcement of those rules, we seem quite divided.
In order to test this further, we polled opinions on various offences. For each of the following videos, we asked if this offence should be punished or not.
As you can see from the results, the majority of respondents find that those offences should be punished, yet in some cases the distance to a 50:50 split is disturbingly close. The answers depend on a particular situation in the clip, the way the question is posed (from viewpoint of the offender versus a potential victim) all the way to the social circle of the respondent (ie. members of the drifting club).
All and all it should be evident from the answers why road rage and vigilantism as a form of crowd based enforcement should not be encouraged.
It is however our hypothesis that a combination of objective and aggregated subjective signals, that erodes away extreme views and normalizes to the interest of public safety, becomes interesting again as a form of crowd enforcement.
What if we
- used machine learning and AI to identify offence videos based on objective measures (ie. crossing red light 2 seconds after it went red)
- used machine learning and AI to anonymize these videos (like we do here)
- had a random group of individuals function as a “micro-jury” to vote whether a particular piece of objective evidence in each video should be processed further or not
- passed only those that go over a certain threshold (ie 80%) to the traditional channels of law enforcement
We call it Traffic Tinder and we quite like the idea! In fact, to the extent that we want to test how many of you want to try being traffic cops for a few minutes regularly? Are you curious what we’ve seen among the first 10 000 offences? Do you want to watch people breaking rules, play the judge and help us evaluate the potential of this method? Well here’s your chance! Sign up with your email below and let’s have a look together!
This isn’t in fact a new idea, but an idea proposed in 2019 by Nick Bostrom from the University of Oxford in his work — The Vulnerable World Hypothesis:
“In theory, some of the monitoring could be crowd-sourced: when suspicious activity is detected by the AI, the video feed is anonymized and sent to a random 100 citizens, whose duty is to watch the feed and vote on whether it warrants further investigation; if at least 10% of them think it does, the (non-anonymized) feed gets forwarded to the authorities”
We think he is on to something!
Towards a better, AI-assisted world!
Silver Keskküla
Founder and CEO
Supervaisor
