I don't quite understand how that works. Surely there wasn't a crowd of 3000 people at the bottom of the escalator when people were walking on the left? Where do the extra people come from?

Public transit in major cities is a set of complex queuing systems. If a station is slower getting people out, they will crowd on the platform and create "back-pressure" on the entire system as the crowd causes incoming trains to take longer to unload, people will see longer queues to enter the system so they'll be marginally more likely to take a cab or abandon their trip.

So some of the 3000 people were literally standing on the platform, but many of them may have been new customers or time-shifted rides that didn't even exist before.

All of this seems like a very micro optimization to eek out efficiency in peak throughput. The bigger macro problem is that transit is centered around sharp peaks due to inflexible work/life schedules. Moving everyone in London or New York around is tough enough, but doing it in a few hours each day is a nightmare and you have to massively overbuild infrastructure.

I used to use the station they trialed this in at rush hour frequently, and over the course of an hour I could believe 3,000 people ended up queueing for escalator space, which is compounded because people are trying to cross the standing queue to get to the right side and walk up. You've also got the flow of traffic from the down escalator doing the same thing.

It sounds crazy, but Holborn really does have vast amounts of traffic in the morning and evening, frequently enough that they'd have to shut the gates until the platforms had cleared enough to fit more people in.