Tesla road trip 2021

Written on 30 August 2021, 09:40pm

Tagged with: ,

See also the previous road trips.

  • Almost 4000km covered this time, a new record
  • Two trips instead of one: to Southern France and to Austria
  • New distance record: 280.3km from Brussels to Metz
  • New range record: only 22kms left in the tank before supercharging
  • No incidents this time 🙂
Brussels to Provence, then Brussels to Austria
The first segment is the biggest – 280.3km. 3784kms in total, excluding the short trips at the destination
280.3kms in one go; still 56km left in the tank…
No risk, no fun 😈
Before the final leg

Non-human football referees – part 2

Written on 19 April 2021, 11:55am

Tagged with: , ,

This is a follow up to the previous article Football is broken. Can it be fixed?

There is a 98 per cent probability that by 2033 human referees will lose their jobs to algorithms.

Source: this study: “The Future of Employment: How susceptible are jobs to computerisation?”
Authors: Carl Benedikt Frey & Michael Osborne
Date: 1 September 2013
Direct link: oxfordmartin.ox.ac.uk/publications/the-future-of-employment/
For more info on how the 21st century will be nothing like the 20th century, I highly recommend reading Homo Deus

How would the players cope with no referee being on the pitch?

Every player knows to ‘play to the whistle’. Other sports are already using buzzers. On top of that, the refereeing decision will be shown on the big screens. The same screens used by the VAR at the moment.

What about the technical difficulties?

The VAR and the GLT are already using a system of cameras. So just add more cameras and some computing power. They are both cheap. On the long run, non-human referees will be cheaper than their human equivalent.

How would that work in practice?

Pre-established rules, multiple cameras, confidence levels, thresholds, continuous learning and fine tuning. These are the HW and SW components of a non-human refereeing system. This is how everything would work in case of a foul:

  • human referees would teach the algorithm what a foul is
  • the algorithm will take the input from the system of cameras and will calculate in real time the probability of a foul. Or, put differently, the confidence level that a given sequence of play represents a foul
  • if that confidence level is lower than a predefined threshold (or in case of advantage play), the play continues. However, if the threshold is reached, the signaling system will kick in:
    • the stadium buzzer will sound and the play stops
    • the decision is automatically shown on the big screen
    • if a yellow/red card is shown, the cautioned player will either acknowledge the caution (if yellow) or leave the field (if red)
    • the system will also project on the pitch the exact spot where the ball will be placed and, if necessary, the minimum wall distance
    • the play will restart with a free kick from the indicated spot
  • the football governing body will fine tune this ‘foul threshold’ continuously (starting with trial matches, but also at the end of the seasons, major competitions, etc)
There are enough money in football to tolerate sub-optimal refereeing