Hacker News new | ask | show | jobs
by NLips 1458 days ago
On the underground you’re meant to stand on the right of escalators to leave room for those walking up on the left. However, TFL ran a trial a few years ago instructing people to stand on both sides during peak times. This actually increased overall throughout and per-person speed: throughput as you can more closely pack passengers standing rather than allowing a smaller number of walkers half the space (and those walkers are necessarily spaced out more than standers), and per-passenger speed because even though walking up the escalator would be faster, the increased throughout reduces the queue time at the escalator entrance, which has a greater effect.

To my knowledge, they didn’t move forwards with keeping the instruction permanent in peak times, presumably because it’s difficult to implement / ingrain in people.

EDIT: link https://www.shortlist.com/news/holborn-escalator-trial-tube-...

14 comments

I've been wondering about that story as well. My own conclusion is that the experiment failed because the metric of average speed per passenger is wrong from a passenger's perspective. People who stand probably don't care about a hardly noticeable increase in speed, whereas people who walk are annoyed by the very noticeable decrease in speed.
I think the speed _for all passengers_ was increased, because of the reduction in queueing.
Their justification for this change, as I read it back then, (the linked article doesn't get into the details) seemed to me to be completely flawed.

It is true that standers are more closely spaced together than walkers. The number quoted was a stander every 2 steps versus a walker every 3 steps. However standers spend roughly twice as much time on the escalator as walkers do. This means that walkers take up about 50% more space*time on the escalator, which is the constrained resource.

This error was exacerbated by station staff observing queue length as a goal to minimize. Walking on the left leads to a longer, and also faster-moving, queue for the left side of the escalator. People are happy to stand in this queue, because they are trying to optimize for the time they get off the escalator, not the time they get on. But station staff sees reducing this queue as a success. This is wrong - firstly they don't take into account the higher throughput for the left side. Secondly, making things less desirable definitely makes queues for them shorter, but this is not a win.

In short, standing-only can reduce total travel time for congested escalators where few people walk (obviously), but it makes things quite a lot worse for busy escalators which are at capacity for both walking and standing. (Even before you account for the fact that walkers on average care more about their speed than standers.)

I lived in London for a year or two and I rarely saw an escalator "at capacity" for walking. You can do all of the napkin math you want, but ultimately, as the article says:

> The one escalator at Holborn that allowed you to walk up the left side during the trial had a total of 115 passengers per minute, but the standing-only ones knocked that up to 151. This was only the case during peak times though – when you’ve got under 100 people on the escalators, it doesn’t really make much difference. Either way, introducing standing-only escalators meant that they only had to SHUT THE GODDAMN GATES AAAHHH once during the trial, whereas it was happening almost daily before.

I have seen lots of escalators at peak capacity for walking. Perhaps you didn't travel a lot in rush hour. People (for obvious reasons) are much more likely to walk up escalators in the morning rush hour than in the evening.

The escalators at Holborn are somewhat exceptional as was noted in the linked article, and the anecdata provided by station staff is not particularly convincing. If the station was extremely congested in the morning, it is hard to imagine that very few people chose to walk straight onto the left side of an escalator rather than queue at length to stand on the right.

'Shutting the goddamn gates' really only happens at stations like Holborn in the evening when people are trying to get out of the city centre. The up escalators at Holborn are conversely at their most busy in the morning, so this is a confusing claim at best.

As for the accusation of 'napkin math', I was simply responding to the clearly wrong napkin math which was widely cited in favor of this change. This purported to show that even at capacity, a walking left side was wasteful of escalator space. Without this flawed claim the main consideration simply becomes that if the left side is being used a lot for walkers it should be kept for walkers (obvious) and if not, not (also obvious).

The trial was at a station with 2 parallel escalators going in the same direction, giving 4 lanes of people. Instead of having 2 standing lanes and 2 walking lanes, they had 3 standing lanes and 1 walking lane. The goal was to reduce queuing for the standing lanes. The walking lanes are usually a lot less busy.
When I travelled through Holborn during this period, they had changed both of the up escalators in the main bank (ticket hall down to Central/Piccadilly fork) to 'standing only'.
It doesn't really work like that, people are lazy and even at peak times will queue at the right hand side to stand, even when the left hand side is completely empty
I got kind of mad at this decision for precisely the same reason as you.

I'd like them to put up some signage to encourage people to walk up (it's a low friction way to get some exercise) and try again. I reckon they could increase throughput and make a few people slightly healthier this way.

One thing I’m conscious of with these crowded escalators is the exit point safety, as some people can be rather slow to move away. Making both sides standing would make me concerned about seeing accidental pileups from people not moving away quickly enough.

You get it sometimes already when the station itself is overcrowded- TfL normally starts limiting access, running congestion control etc. Fortunately most of busier places (Bank, Canary Wharf, etc) are full of regular workers who know the drill.

It’s more one off special events where causal travellers show up en masse.

Just want to comment that, that's essentially public transit within a public transit. Public transit cares for throughput, private transport (cars, left-lane of escalator) cares for latency. Queuing theory in a nutshell.
In Moscow metro they sometimes ask (but do not enforce) via speakers to stand on both sides of the escalator. I heard this practice was brought by McKinsey from London. In my experience this sometimes works on very short escalators, and on the longer ones people seem to ignore the suggestions. I wonder about the psychological reasons of this behavior.
I took part in this trial, at Holborn Station. The escalators are very long, and few people walk. However, despite the signage, markings and announcements, when I stood on the left I still got the occasional tirade of abuse from someone attempting to walk up behind me! Most people continued to only stand on the right as a result.
> On the underground you’re meant to stand on the right of escalators to leave room for those walking up on the left.

Don’t they drive on the left in the UK? The “slow” lane should be on that end.

Turns out it’s just legacy from the original escalator design.

“It all dates back to when the first successful escalator (there was a failed attempt to build a spiral escalator beforehand) was introduced on the tube in 1911 at Earl's Court Station. Alighting an escalator was different back then to nowadays. There was a diagonal partition — beneath which the stairs disappeared — that shunted passengers disembarking to the left. So it was decided that those walking up the escalator should stand on the left, otherwise they'd have to cut through a line of those standing. And that would've been mayhem.”

https://londonist.com/london/transport/why-don-t-we-stand-on...

We drive on the left, which means we walk on the right (to see traffic coming), so I guess it stems from the latter
In Australia, we drive on the left, but on escalators we pass on the right just as the parent comment envisioned.
In the US, we walk at traffic when there are cars on the road but the passing lane is the same (left) for driving roads and for pedestrian multi-lane "roads" like escalators.
This was also only at Holborn, one of the typically most overcrowded stations on the network.

A lot of the problem was that it was relatively unpopular, even with the increase in throughput. There was some talk about having at least one walking side per set of escalators, but that's also harder from a messaging point-of-view.

> This actually increased overall throughout and per-person speed:

Increasing the average per person speed sounds like the expected result. Not necessarily a net win if the people sufficiently motivated to exit quickly to walk (who tend to bypass queues) are slowed though.

They still had the option of walking though. With two parallel escalators, 3 of the 4 "lanes" were for standing and one for walking, rather than reserving two lanes for walking when few people do so.
The study shows this was better at peak times, especially for the goal of making sure there are no queues at the bottom of the escalators which can slow the tube down, but presumably in normal times (when there is space for standing and walking) this is worse for individuals who have a preference on walking or standing.

I get the tube to and from Holborn every day and there are always people (me included) walking and I rarely arrive at peak times.

The way I understand it is that impeding someone trying to pass on an escalator in London (it doesn't have to be on the underground) is socially unacceptable. You can actually stand on whatever side you want, but you'd better be ready to get out the way if you decide to stand on the left.

Making everybody stand makes people uncomfortable because it's essentially forcing them to inconvenience other people.

I've always thought of it as a classic latency vs throughput tradeoff. You sacrifice some individual latency (of the people on the left walking quickly) for improving the overall throughput of the system (and therefore worst case latency).
This is the way I remember it in Tokyo: stand on the side, unless it's super busy. Then you jam-pack as many people as humanly possible.
'Stand on the right' is relative to which way you are facing. You would still be correctly following the instruction if you are facing opposite to the direction of travel and standing in the way of 'walkers' (on your right). I find Londoners generally receive this sort of humour very poorly ;)