Mathematical model suggests London Underground may be 'too fast' | Hacker News Reader

Mathematical model suggests London Underground may be 'too fast'

bbc.com

32 pointsuxhacker28 comments

The ideas and models are interesting from a systems thinking perspective and it is certainly true that local optimisations can have unintended systems consequences, but unfortunately it doesn't model reality when it comes to London.

The paper makes the flawed assumption that people commuting into London from outside the boundaries of the Underground network start their journey by driving to the boundary and entering the system at that point, which as anyone in London can tell you is not the case. Most commuters from outside the network will arrive at one of the main overland train terminals (Euston, Paddington, Waterloo, Victoria, Kings Cross, etc.) and change mode of transport at those points. Those points certainly do see large congestion, but they are a very different beast to road traffic so I very much doubt the same conclusion would be reached if they were included in the model. The paper also ignores the bus network, which is also heavily used in London, which is I think also different to say New York.

If you are going to start studying systems then you need to look at the whole system otherwise you are still just doing 'local' optimisations. If their conclusion/recommendation were to be implemented, I suspect it would make things worse overall as they haven't understood the system.

This article suffers from serious Chinese whispers.

The clickbait title doesn't reflect the conclusions of the article, which don't reflect the conclusions of the paper, which don't reflect the actual evidence given by the simplistic experiment that the physicist responsible for it decided to use to model the real-world system.

The guy who wrote the paper was doing some sensible academic research that really should have had no conclusions - it just worked towards a new facet that should be used when modelling transport systems in cities, which is perfectly commendable and useful, but can't be relied on as a standalone system (in the same way that you couldn't model all of the interactions in the universe using only magnetism). Deciding to draw the conclusion that he could predict an optimal speed for the underground to move at was foolish and detracts from the original paper (although I'm sure it garnered it much more attention).

That the guy is quoted at the end of the article admitting that his conclusions are evidently nonsense and that they should probably throw in some "secondary terms to account for complications he just thought of" is almost too perfect.

https://xkcd.com/793/

I have to admit I was wondering if at some point the author was modeling passengers as perfectly spherical objects vibrating in a sinusoidal fashion.

It reminded me of all of those traffic studies where they conclude that they can increase traffic flow through a city by reducing all of the speed limits and taking away lanes and then something magical happens. Or the counter studies that say adding lanes reduces the total volume of traffic a road can process because it will make more people take the road and cause it to back up.

> "We create these connections, and then we make an assumption, which is: When someone wants to go from A to B, they look for the quickest path - whatever the mode."

That sounds like a crazy assumption for commuting.

This may be true for some subset of commuters, but when commuting via public transport regularly, a lot of people quickly develop favoured routes that often have nothing to do with speed, and everything to do with flow and convenience and what trains you can get seats on, and what lets you actually spend your time reading rather than changing trains.

Over 15 years of commuting in London, the quickest path has almost never been the route I ended up favouring.

It's true. How much time is non-stop worth? Or a seat the whole ride? It is worth something!

I've been willing to about doubling my commute to be able to just sit down on a bus or train the whole way. I treated it as a time saving: Yes, I spent more time commuting, but that time was spent napping or reading, so it was a net gain for me in usable time.

It's also a cost issue for many. An annual Zone 1-2 Travelcard costs GBP 1280,-. A Zone 1-9 Travelcard costs GBP 3336,-. The cheapest single zone annual cards cost 964,-. While a Bus and Tram pass costs 840,-.

Depending on where and how often you need to travel, for many it's worth it to pay the time penalty of taking the bus rather than paying for one or more extra zones, or paying for tube/train at all.

It's not that uncommon for people to do 1-1.5 hour bus commutes that'd have been far shorter, and more expensive, by tube/train.

So you are occupying a seat/space on a train longer than you absolutely need to? You are therefore adding passenger-miles to the system, reducing availability for others.

I don't suggest ill intent but if people are using anything other than the most efficient route, the route that gets them out of the system at their destination most quickly, that is an inefficiency to address.

This analysis ignores people who use only the train and tube or just the tube. If you hinder the tube to try and improve traffic what will happen is traffic will eventually increase where you have made improvements (the bottlenecks referred to in the article) and the tube service has still been degraded.

The failure of the analysis is to assume that car drivers should be a consideration (perhaps that is due to the US bias toward car travel over public transport). Really we need to be improving public transport such as the tube so much that car driving is disincentivising as much as possible and public transport is incentivised as much as possible.

I'm inclined to question whether the study authors have ever been to London, never mind examined actual passenger throughput data. Most tube stations have far too little parking in the vicinity to create a "bottleneck" by encouraging overuse of personal transport, and "bottlenecks" with bus services are far more sensibly resolved by increasing the frequency of bus services at peak times than slowing trains down.

I think you're right on your first paragraph

But the consideration about traffic is not only about cars but people that do Bus->Tube and vice-versa

I can't see any part of the study that distinguishes between the congestion effects caused by cars and buses (or indeed walking and cycling, which the naive speed-optimisation specification of the study assumes away altogether)

This is pretty critical to some of the conclusions: a subway line creates congestion at the ends of subway lines rather than them becoming a logical nexus for bus routes.

True that article ignores passengers using only tubes or trains. But I think the article suggests global congestion may be improved by slowing down certain parts of a system.

Even considering only people using just the tube, the tube station can switch only a finite number of passengers in and out of tubes. If a system has two bottlenecks, improving the narrower bottleneck will result global improvement only until that bottleneck is no longer the narrowest.

It is possible that increasing tube throughput would increase arrival rates at the destination stations. And if we consider congestion as nonlinear to number of passengers arriving at the station, then this might decrease the throughput at that station (beyond some threshold arrival rate); (I think this transition is similar to a phase change). That would mean the second bottleneck would effectively becoming slightly narrower in capacity, and thus the global throughput decreases.

The trick would be to improve the narrowest bottleneck; but not too much, that it would cause such phase transitions at other points in the network. At which point, we need to start looking at other bottlenecks.

This is an interesting study, but the "conclusion" is a bit odd. To me, if an efficient railroad system is causing congestion near the "park and ride" or equivalent, the answer is to expand the rail network further not to artificially handicap it.

The problem is that models rarely replicate reality for very long. I live in a particularly troublesome bit of London and they used some mathematical analysis to optimise traffic light timings in the area and celebrated that it was as success. That was premature; it's measurably worse within two weeks after people realised that it was a better route and promptly clogged it up.

> Of every well-meant proposal, ecologists ask a standard question: 'And then what?'

-- Garrett Hardin

That's the nature of traffic. You make a highway twice as wide and it's great for a few years until enough people go "Oh! I can live twice as far away, have much lower rent, and still keep my commute under 2 hours!"

And then you end up with LA.

If anyone's curious to read more about this, Wired has a pretty good piece: http://www.wired.com/2014/06/wuwt-traffic-induced-demand/

Slowing down trains to solve a bottleneck downstream is like squeezing the bump in the carpet to another location in the room.

It doesn't make anyone get anywhere any faster, and hurts the commute times of the people who are not involved in the bottleneck (because, say, they walk or bicycle to their destination from the train station, or take some other fast train).

Transportation is about optimizing the average trip time; concern about accumulation at transfer points is secondary.

This analysis seems based on preventing congestion. But congestion, while ugly, isn't a bad thing. What matters is total journey time.

Imagine if you replaced all the trains with trek-style teleporters. Rather than 100 people all getting onto a train and being whisked away, they would all wait in a line outside these devices. That would be an increase in congestion, more people standing around waiting rather than moving. But if the teleporters get them to their destinations faster nobody cares that they had to wait in line.

It's a matter of perspective. The city/transport planners see people standing around not moving as a bad thing. But the people don't care. Standing on the slow train or standing in the line waiting for the fast train is all the same to them. What works best for the people isn't the an elegant system of slow trains to ensures nobody ever stops, where nobody ever suffers the evil congestion. What works best for them is getting to work as fast as possible. If that means crowds at bottleneck where in-feed is slightly faster then out-feed so be it. Slowing things down, having more people spend more time standing in trains rather than on platforms, doesn't help them get to work faster.

Would it even be possible to slow down the Tube while maintaining capacity?

You would have to either make the trains bigger (not possible on most lines), or add additional trains - expensive, and potentially a safety issue as you need to make sure there is always an evacuation route available.

Some lines already operate with more trains in service at a given time than there are platforms on the line.

This study sounds suspicious to me. I wouldn't be surprised if it were sponsored by cab companies, bus companies, and bike activists :)

Does this model assume everyone in London has a car?

No. But it does assume that some people can choose to go part of a trip by car (their own or another) and the rest on the tube, and that these mixed-mode trips affect congestion for everyone.

But thats only going to happen at the extreme ends of the tube system no one is going to drive further into London to park and then get the tube

Quoting the article: "key locations outside the city centre, where people switch transport modes, become bottlenecks" and "the underground network, for example, tends to decrease congestion centrally but increase it where the underground lines finish."

From experience The bottle necks in the tube are not the extremities like watford or finchly.

No, but the problem is that it assumes that cars are as important as public transport which can't be further from the truth.

Driving in central London ROTFL

So if the goal of London Underground were to avoid street congestion, and changing the way the street and underground networks meet were impossible or illegal, then lower speed would be appropriate.

That's interesting. In an if-and-if sort of way.

Reply on news.ycombinator.com