So in the concept video it shows passengers loading on to a sled at surface level that can hold maybe a dozen people, and then it descends on an elevator 50 metres to the tube and goes down the line. This sounds cost effective compared to a full on underground subway station, which really aren't cheap to build even if you 10x the cost of tunneling, but it isn't good for much unless all those passengers intend to also get off at the same stop, which isn't really something that happens. So if one of those dozen passengers wants to get off, then all 12 will have to go all the way back up to the surface and back down again, which could end up being cripplingly slow.
I think you're right that stops will be longer, but it's my understanding they'll also be less frequent and they'll cover greater distances than a regular subway track (in say, NY). So you'll end up getting there faster and not have the stress of driving in traffic.
For a conventional subway at rush hour, there's 1000s of people getting on or off every hour at any given station, which isn't feasible if you're only loading a dozen people at a time. The alternative being to build proper underground stations, which are a big part of that 'billion dollars per mile' subway construction cost.
And this reminds me of Musk's Hyperloop whitepaper, where he estimated it would cost 10 billion to run an evacuated tube from LA to SF, which was ridiculous. There's all this reality that he failed to factor in to his calculations, and once you've addressed the full costs relative to people moving capacity, all of the sudden Hyperloop doesn't look so great compared to conventional high speed rail, and his Boring concept starts looking more and more like conventional subway.
Elevator engineers sometimes handle this problem by having "express" elevators that travel between distant floors and "local" elevators that operate between nearby floors.
That avoids the need for big central stations (other than the ground-floor lobby, of course, but there's no way around that for a building -- a street transport system isn't limited to a single point of entry).
With intended throughput and few stops, allocating in bunches of 12 people with the same destinations or at least least stops automatically is trivial.
