This is now a completely different topic. It probably contains many over-simplifications, probably a lot of wrong facts and no clear evidence. Just thoughts. So, add your grain of salt to the following sentences.

For a while, I thought about a very simple question: “Are traffic light broken by design?”. I’m riding a bicycle also in the city and it seems that every junction has different rules (sometimes the traffic light applies to the bike lane as well, sometimes the bike line has own traffic light, sometimes the bike line disappears just before the junction and sometimes there are no bike lanes at all). However, almost always, there is a problem when the cars want to turn right, but the cyclists don’t. That’s a conflict. You can extend this to pedestrians: If they want to go straight, but the car wants to turn right, this is out of control of the traffic light.

What’s the purpose of a traffic light?

From my understanding, the main goal of a traffic light is to prevent accidents. And the second goal is maybe to optimize traffic flow. How do you prevent accidents? You make sure, that there is no possibility, that traffic participants from different directions can cross each other. Easy.

This is possible with a simple traffic light, if the junction is simple. That means, if there is no way to turn left or right. Then you can allow traffic either on the horizontal direction or the vertical direction, as longs as you don’t allow both at the same time.

This “simple” junction of course doesn’t exist in reality. This junction doesn’t make sense - in fact, you don’t need a traffic light at all, if you build a bridge. Then the two directions can flow in parallel, because they don’t touch each other anymore.

In reality however, you want to be able to switch the directions, turn right or left. Turning right is still simple, this can be always allowed, as you don’t need to cross any lanes for this. Note: I’m talking here about the situation, where you drive on the right side of the road…
Turning left is different: You need to let pass the oncoming traffic before you go - although your traffic light showed green. As this might be dangerous, other special traffic light controls have been invented, like a green arrow light, which indicates that the oncoming traffic has to stop (it sees a red signal) and you can assume, that there is no oncoming traffic.

Ok, but that’s still not the real reality: In cities, there are also pedestrians, that also want to cross a junction and cyclists. Usually, these are positioned right to the rightmost driving lane. And this causes a problem now, if a car want to turn right, but it needs to let pass any pedestrian or cyclist on the right. And here all traffic participants have a green light from the traffic light and are good to go, but still are not: The traffic light doesn’t fulfill its simple goal once again: prevent accidents. To prevent such possible accidents, you would need to introduce separate phases, where e.g. no car is allowed to turn right while pedestrians are crossing the street.

Does it achieve its second goal - optimize traffic flow? Maybe: in case, there are no pedestrians or cyclists, then the cars don’t need to wait. If there would be a separate phase for pedestrians, then they would always need to wait. In that sense, maybe the cars can leave the junction behind in some cases a bit faster.

It raises other questions: For whom are traffic lights optimized? For whom are cities at all optimized? This leads to the science fictional joke: Imagine aliens are visiting the earth. Which life form would they observe to be the primary life form on planet earth? The answer is: cars. Which is only slowly changing in the last years with ideas like superblocks in Barcelona.

Could it be different?

If we consider the first goal of a traffic light - to prevent accidents - the most important goal, then we always need to demand separate signals for left- and right-turners. That definitively would make the traffic flow slower, as you need for each direction an own phase, where no other participants are allowed. And maybe you need additional extra lanes, but that’s already an optimization.

Another idea I had (and I’m not surprised, that I’m not the only one having this idea): What if we separate the cars, which are usually very fast, from the pedestrians, who are usually very slow? Which means, while pedestrians are crossing the street, no cars are allowed. That should prevent any accident between cars and pedestrians. In fact, we could even allow the pedestrians to cross in any direction, if no cars are allowed at the same time. Even allow them to cross diagonally.

This is known in German as Rundum-Grün - green in all directions (for pedestrians). Additionally crossing diagonally is called Pedestrian scramble.

A simplified light signal control program could have the following phases:

  1. allow only the pedestrians.
  2. allow the cars in horizontal flow, including left- and right-turners.
  3. allow the cars in vertical flow, including left- and right-turners.

This adds just one additional phase for the benefit of preventing accidents. It would be interesting to compare junctions with this “new” kind of control with junctions using the ordinary control in terms of traffic flow. I would not be surprised, if the green in all directions would be faster overall, as with the ordinary traffic lights, the right turners often need to wait for pedestrians and the traffic is blocked - in extreme cases so much, that only one car can make it before the light is red again.

I don’t know, how this would affect cyclists. If they should cross together with the pedestrians, when all directions have green or if they better should cross together with the cars. The conflict between cars and pedestrians (cars fast, pedestrians slow) appears again, just between cyclists (fast) and pedestrians (slow). Although the difference in velocity is not so high. It also depends on how the junction is built: If there are bike lanes separate from the (car) driving lanes, then you can’t allow cars and cyclists having green at the same time, as the cars would then need to cross the bike lane.

In an ideal world, the different conflicting types of traffic participants wouldn’t need to share the same infrastructure. E.g. on highways, no pedestrians are allowed - for good reasons. However, in cities, pedestrians and cars share the same street. And cities tend to be built for cars only (not just in the US), but humans are living in the cities, not cars… If you cycle through a city, you sometimes feel like you are in a huge car park, as most of the cars are parking anyway.

A more radical alternative (probably not realistic) would be: no traffic light at all. You are only allowed to cross the junction in walking pace - that means: cars need to be super slow, and cyclists as well. Then the idea would be, that it is self organizing. All traffic participants look after each other and coordinate with each other, who goes first.

New experiments (in Munich)

The interesting thing is now, that in Munich the green in all direction traffic light is tested at least on one junction: Pilotprojekt „Rundum Grün“ im südlichen Bahnhofsviertel.

It’s the junction “Landwehrstraße / Goethestraße”. You can also have a look at street view: https://maps.app.goo.gl/LbPLK9jEk6WGH4jM9.

This is an experiment with a duration of one year. It is accompanied by studies, so that we can observe and compare the changes in traffic flow before and after. The overall goal is to increase the traffic safety for pedestrians. I’m excited to hear about the results of this experiment.

However, it is only a small junction, in the sense that the horizontal traffic flow is one way (which makes turning left easier), and there are no separate bike lanes. That means, it’s just tackling the conflict between cars and pedestrians. I’m not sure if these results can be used for other, bigger junctions. But at least, there is a beginning.

In April 2024 this year, a deadly accident happened at a bigger junction in Munich: Verkehrsunfall zwischen Lkw und Fahrradfahrerin – eine Person tödlich verletzt – Trudering. It involved a cyclist and a truck.

The junction is here on Openstreetmap: https://www.openstreetmap.org/#map=19/48.121420/11.664691 As you can see on street view (https://maps.app.goo.gl/dTnVvACdi5zRCqmz6) it contains an extra driving lane for right turners, but it also contains an extra bike line. If a car or truck wants to switch to the right lane, it needs to cross the bike lane - which makes this very dangerous for cyclists. Extra bike lanes sometimes help to increase safety (especially if they are completely separated), but sometimes they are counterproductive. Especially on bigger roads, like that one, which as 2 driving lanes in each direction plus bike lanes and sidewalks.

Conclusion

And I’m happy to see, that alternative solutions are at least being tested. Let’s hope, that this results in better safety for all traffic participants, whether they are using just their feet, or using a bicycle or a car. And maybe, this can be an enabler, to revise other junctions as well.

On the other hand, I’m a bit surprised, that there is no clear scientific driven answer to these safety question yet - as why would we need to do an experiment then? I did a short search for traffic simulators - as simulating would seem to be one approach and there are simulators available. E.g. Microsimulation of Traffic Flow. This one however seems to be forgetting pedestrians…