> For this study, the Google Maps algorithm was modified to prefer alternative routes with similar travel times and segment types, effectively guiding trips away from the pre-selected congested segments
> Over a six month period, we adopted a city-wide switchback (also known as crossover) experimental design, alternating between this treatment and the control (unaltered) routing algorithm over consecutive days to appropriately measure the effect of this intervention
> Averaged across cities, we observe a median increase of around 2% in driving speeds on targeted segments, corresponding to a median decrease of 0.5% to 1.0% in fuel consumption rates
The cities were: Atlanta, Boston, Chicago, Los Angeles, Miami, New York, Philadelphia, Salt Lake City, San Francisco, Seattle.
One of the things this type of intervention doesn't take into account is that different roads are built with different levels of hardiness based on the amount of traffic they are expected to receive.
For instance, a few years ago, a segment of I-495 in Delaware needed to be unexpectedly shut down for emergency repairs. Drivers were rerouted. But because of the increase in traffic on the less-hardy detour route, that route needed repairs and repaving soon afterward, much more quickly than it would have ordinarily required.
So yes, drivers can be better dispersed to ease congestion, but we also need to consider the secondary effects to the roadways themselves.
A random factoid that will always stick in my brain is that the damage a vehicle does to a road is proportional to the weight on an axel...to the fourth power. I suspect rerouting semis to the roadway is by far the greatest factor.
Traffic apps only know about congestion if someone running the app goes down the congested road. Because of this, I've always suspected that the apps, from time to time, will route someone down a route they haven't gathered data on in a while, just to collect the data, and even if the route is likely to be suboptimal.
>I've always suspected that the apps, from time to time, will route someone down a route they haven't gathered data on in a while, just to collect the data, and even if the route is likely to be suboptimal.
I can't say this actually happened for me. For straightforward routes with no congestion I never saw random alternate routes being proposed. That makes sense, given that it'd probably tip people off. If this is happening, they must only be doing in cases where there's congestion and the difference is marginal, eg. it's rush hour and the "optimal" route takes 30 minutes but the alternate takes 33 minutes. Moreover you don't really need any deliberate effort to see this effect. If nobody is traveling on a side road, the algorithm will probably revert to historical patterns, which might turn out to be overly optimistic in congestion scenarios (eg. there's nearby road repairs and other people are already using it as an alternate), thereby giving you the impression that you got screwed over by the app.
Doesn't have to be true. My state has public camera feeds for various highway stretches. Too low-detail to identify vehicles but easily enough to detect congestion.
I wonder this same thing when I’ve been routed completely asinine ways off interstates at night when traffic appears calm. I always have to play the game of “is there horrendous traffic ahead from night paving / accident, or am I being conscripted as a traffic sensor robot?”
When Google Maps routes me using a smaller secondary road instead of the main road that I would otherwise have used , I've always wondered whether that significantly changes the amount of traffic that smaller road sees. It's funny to consider that arbitrary black-box changes to the routing algorithm can have a noticeable effect to people that live there.
If we could copy the traffic laws of a country like Germany to the US, I think that would have the biggest + cheapest effect. I am OK with automated/elevated enforcement if it means stop & go traffic evaporates into free flowing conditions.
We should also take the idea from Finland where the traffic fines scale with each person's ability to pay. $100 for camping the passing lane or failing to maintain a reasonable following distance is not a big consequence for a lot of people. $100k covers the edges a lot better.
>During “treatment” days, the modified routing guided all trips that encountered the pre-selected congested segments toward alternative routes with similar travel times.
Why would they specifically need to route people away from congested segments? Presumably if a segment gets congested enough, it'd be considered slower and therefore won't get picked in the first place?
Every actor acting rationally in their own interest can get to a worse equilibrium than if they were coordinated(in this case, to completely ignore the new edge). There are many other examples of this in game theory, you should look it up.
(I'm not saying I'm the first or only person to think of it, but I did patent it, as well as the extra claim of compensating those people sent on the slower route).
For over a decade, I imagined that if I ever landed a job at Google, this would be my most significant project. It made me chuckle a bit when I read the announcement, they finally built it! confirming that my thoughts weren’t entirely delusional XD
The way to maximise traffic congestion would be to remove trucks off the road during peak times, banning trucks from the fast lane at all times, making it the social norm to toot someone in the fast lane so they move over.
I have a lot of difficulty thinking those changes alone would solve congestion. I’m especially thinking is ski traffic along I70 from Denver to the mountains. The issue is just volume of passenger vehicles, and there are no other realistic routes
> For this study, the Google Maps algorithm was modified to prefer alternative routes with similar travel times and segment types, effectively guiding trips away from the pre-selected congested segments
> Over a six month period, we adopted a city-wide switchback (also known as crossover) experimental design, alternating between this treatment and the control (unaltered) routing algorithm over consecutive days to appropriately measure the effect of this intervention
> Averaged across cities, we observe a median increase of around 2% in driving speeds on targeted segments, corresponding to a median decrease of 0.5% to 1.0% in fuel consumption rates
The cities were: Atlanta, Boston, Chicago, Los Angeles, Miami, New York, Philadelphia, Salt Lake City, San Francisco, Seattle.
The data and code is also available (https://github.com/google-research/google-research/blob/mast...) from the paper (https://www.nature.com/articles/s44284-026-00443-x).
Kudos Google! Nice to see this kind of work. That said, let's just build more trains?
One of the things this type of intervention doesn't take into account is that different roads are built with different levels of hardiness based on the amount of traffic they are expected to receive.
For instance, a few years ago, a segment of I-495 in Delaware needed to be unexpectedly shut down for emergency repairs. Drivers were rerouted. But because of the increase in traffic on the less-hardy detour route, that route needed repairs and repaving soon afterward, much more quickly than it would have ordinarily required.
So yes, drivers can be better dispersed to ease congestion, but we also need to consider the secondary effects to the roadways themselves.
A random factoid that will always stick in my brain is that the damage a vehicle does to a road is proportional to the weight on an axel...to the fourth power. I suspect rerouting semis to the roadway is by far the greatest factor.
Traffic apps only know about congestion if someone running the app goes down the congested road. Because of this, I've always suspected that the apps, from time to time, will route someone down a route they haven't gathered data on in a while, just to collect the data, and even if the route is likely to be suboptimal.
>I've always suspected that the apps, from time to time, will route someone down a route they haven't gathered data on in a while, just to collect the data, and even if the route is likely to be suboptimal.
I can't say this actually happened for me. For straightforward routes with no congestion I never saw random alternate routes being proposed. That makes sense, given that it'd probably tip people off. If this is happening, they must only be doing in cases where there's congestion and the difference is marginal, eg. it's rush hour and the "optimal" route takes 30 minutes but the alternate takes 33 minutes. Moreover you don't really need any deliberate effort to see this effect. If nobody is traveling on a side road, the algorithm will probably revert to historical patterns, which might turn out to be overly optimistic in congestion scenarios (eg. there's nearby road repairs and other people are already using it as an alternate), thereby giving you the impression that you got screwed over by the app.
Doesn't have to be true. My state has public camera feeds for various highway stretches. Too low-detail to identify vehicles but easily enough to detect congestion.
I wonder this same thing when I’ve been routed completely asinine ways off interstates at night when traffic appears calm. I always have to play the game of “is there horrendous traffic ahead from night paving / accident, or am I being conscripted as a traffic sensor robot?”
Government data sources can input real time data into Google Maps [1]. So, it's more likely than not that local police or state DoT is behind that.
[1] https://contentpartners.maps.google.com/
When Google Maps routes me using a smaller secondary road instead of the main road that I would otherwise have used , I've always wondered whether that significantly changes the amount of traffic that smaller road sees. It's funny to consider that arbitrary black-box changes to the routing algorithm can have a noticeable effect to people that live there.
If we could copy the traffic laws of a country like Germany to the US, I think that would have the biggest + cheapest effect. I am OK with automated/elevated enforcement if it means stop & go traffic evaporates into free flowing conditions.
We should also take the idea from Finland where the traffic fines scale with each person's ability to pay. $100 for camping the passing lane or failing to maintain a reasonable following distance is not a big consequence for a lot of people. $100k covers the edges a lot better.
https://www.theguardian.com/world/2023/jun/06/finnish-busine...
Where isn't there traffic congestion?
Interesting, I always just assumed they were already doing this.
>During “treatment” days, the modified routing guided all trips that encountered the pre-selected congested segments toward alternative routes with similar travel times.
Why would they specifically need to route people away from congested segments? Presumably if a segment gets congested enough, it'd be considered slower and therefore won't get picked in the first place?
Probably many reasons but here’s one demo: https://en.wikipedia.org/wiki/Braess%27s_paradox
Every actor acting rationally in their own interest can get to a worse equilibrium than if they were coordinated(in this case, to completely ignore the new edge). There are many other examples of this in game theory, you should look it up.
Hey I patented this idea about a decade ago:
https://patents.google.com/patent/US20170184409A1
(I'm not saying I'm the first or only person to think of it, but I did patent it, as well as the extra claim of compensating those people sent on the slower route).
For over a decade, I imagined that if I ever landed a job at Google, this would be my most significant project. It made me chuckle a bit when I read the announcement, they finally built it! confirming that my thoughts weren’t entirely delusional XD
The way to maximise traffic congestion would be to remove trucks off the road during peak times, banning trucks from the fast lane at all times, making it the social norm to toot someone in the fast lane so they move over.
I have a lot of difficulty thinking those changes alone would solve congestion. I’m especially thinking is ski traffic along I70 from Denver to the mountains. The issue is just volume of passenger vehicles, and there are no other realistic routes