Latency, Throughput, and Walking on Escalators

blog.tacertain.com

23 points·hakejam··3 comments

There's a middle ground here that the author is missing:

It's not that walking on the escalator slows it's throughput, but that a walking lane on the escalator slows it's throughput.

Thus the best throughput is obtained when you go with a policy of stay right if possible but no walking lane. Under heavy load this degrades to standing, under light load it permits people to move faster.

Yeah, I've never encountered the social behavior the author describes. In my experience, people walk when the escalator happens to be clear ahead of them, and stop behind the next cluster of standers. I've never seen standers make an effort to stay to one side. So walking is just an opportunistic and limited benefit to the system. Perhaps its regional.

in East Asia, it's entirely normal for one side of the the escalator to be full of people standing, and there's a long queue at the bottom to join the standing side, while the walking side is almost empty, often entirely empty if it's a long escalator. "just in case" someone wants to walk, and it would be a terrible social faux pas to be standing in someone's way, on the walking side.

Saitama Prefecture is trying to change this behaviour, get people to stand on both sides, and the reaction from the rest of the country has been very mixed, a lot of people are making fun of Saitama people as having nowhere to go in a hurry. It's very inefficient, and the culture will take a lot of time to change.

Reply on news.ycombinator.com