Hacker News Reader

Comment by cs702 | Hacker News Reader

In addition to the correlation found by these researchers, I suspect the Braess Paradox[1] and other complex network phenomena are also at work.

[1] Adding more roads to a network of roads can make traffic worse (and viceversa, closing roads can make traffic better). See http://en.wikipedia.org/wiki/Braess's_paradox or watch this friendly explanation: http://www.youtube.com/watch?v=ZiauQXIKs3U

Edits: moved links and parenthesis to footnote, and added "In addition to the correlation found by these researchers" and "other complex network phenomena" to make my point clearer.

View on HN

This is not the same statement. If I have one path which has a certain amount of bandwidth and subsequently double it (making the road twice as large), the overall speed of the system must either increase or stay the same (the latter if the system is under capacity).

That's the paradox! The difference between a traffic route and a bandwidth-pure pipe (apologies for idiosyncratic terminology, i'm not a trained traffic planner) is that cars will choose their routes selfishly. The way I think of this difference is that even on a straight road that doubles its number of lanes, and we've all seen this, cars will change lanes with the idea that they'll get through traffic. Unfortunately, this lane changing, especially in gridlock situations, slows other cars down more than their overall speed is increased. Thus the overall bandwidth of the road, such as it is, is reduced. Add brake ripples and slow reaction time when moving in bumper-to-bumper, and the reduced overall speed should be obvious.

The real problem is that the roads don't seem to be engineered with a "bandwidth" mentality. Instead I see interchanges where 2-3 lanes from a new road are being piped into an existing 2-3 lane road. Did the existing road get expanded to 4-6 lanes? Very rarely, because you would see a lot more 10-20 lane roads running through downtown.

Instead, it seems a lot of "traffic" studies are just minor tweaks around the edge, add a lane or remove a lane from a 4-6 lane main corridor and the traffic patterns basically don't change that much because its massively oversubscribed already.

actually they manipulate the flow of cars to the smaller roads by making u wait on the offramp longer when its busy iirc, after watching a few docu's on the engineering i think theyre engineers are really good and bright

the book "selfish routing and the price of anarchy" by tim roughgarden is about this. its really interesting for its own sake, though i dont know how relevant it is to this particular thread's article

The video doesn't do a good job of explaining this at all.

Ultimately the reason why adding more roads to a system produces more traffic is more people take cars vs. public transit.

No, its much more complex than that.

Adding road capacity makes other types of trips possible. It's not a simple "hey, I'm taking the bus today because they built a lane". It's more like "Hey, I don't need to worry about uprooting my family or spending 6 hours on the bus to get a job in place X vs Y."

Think about it in computer networking terms. Netflix and Office 365 didn't exist in 1995 because it wasn't possible to make those sorts of solutions work then. Today, we need to worry about network congestion because millions of people are streaming movies.

Very much so. New and bigger roads cause all kinds of associated growth. New businesses and home pop up because it's now practical to get to them in those locations. This increases demand for the roads, so they eventually become congested again. But not just because people like to go from place to place for no reason other than to go.

It would seem then one way to spur economic growth would be to build more roads, transit, etc. Just as building faster internet spurs more internet companies.

Been there, done that... it's called 1946-2006.

It's probably the same reason why we just get bigger hard-drives instead of cleaning stuff we store in our current storage capacity.

> Ultimately the reason why adding more roads to a system produces more traffic is more people take cars vs. public transit.

More precisely, more people take cars vs. not take cars. "Not take cars" can be public transit, human powered transport or even not moving at all.

This correction might seem pedantic, but I think it is important. One of the lessons of these "paradoxes" (more roads = more congestion) is that one must take into account the whole traffic system.

I'd add that another salient factor is people's housing choices relative to jobs and other destinations -- there is an interplay between road-building and housing construction.

No the video is about the The Braess Paradox.

You're talking about the Downs–Thomson paradox. that's a related, but different, problem.

[1] https://en.wikipedia.org/wiki/Braess%27s_paradox [2] https://en.wikipedia.org/wiki/Downs%E2%80%93Thomson_paradox

That's in addition to leading people to move further out -- which also intensifies the traffic at the re-entry points to the thinner roads at the center, which are march harder to increase in capacity.

bigdubs: that video explains the Braess paradox, not the correlation found by the researchers. I modified my comment to make that clearer. Thanks!

Please explain how he can include a route to switch between routes that takes zero time. That does not make logical sense at all. It kind of ruined any premise he might have started with.

A common example of this is an intersection. But even without that example, I don't know how that could ruin the explanation. Zero switching time is just used to simplify the analysis, the overall phenomenon should be clear.

I think that was just to keep the math easy. The principle holds if the switching time is non-zero, but the math becomes less clear.

The principle only holds if you are comparing switching between north-south or south-north. If you compare switching to not switching, it doesn't hold.

Let the switch cost something like 1 unit and the analysis on the Wikipedia example still holds - at the switch, drivers face a choice between a 41 minute route and a 45 minute route. Yes, this paradox is a property of the specifics of the network in question, but it may hold for real networks is the claim.

Reply on news.ycombinator.com