3 reasons why Uber can’t replace transit

Transit projects from Detroit to Nashville are running up against a new argument from opponents. The latest line from anti-transit types is that ride-hailing apps like Uber and Lyft are going to make fixed-route bus or rail service obsolete.

It doesn’t hold up if you’ve given some thought to the huge amount of space cars consume compared to buses or trains. But many people don’t spend their days thinking about the spatial efficiency of transit.

If you find yourself arguing with someone about why transit is essential, a new fact sheet from TransitCenter can help. It makes the case that Uber can’t replace transit in clear, compelling terms that anyone can understand [PDF].

Here are TransitCenter’s three points for transit advocates to commit to memory.

1. Uber and Lyft hog too much space

Let’s say, hypothetically, that a city gives up on transit service because officials think Uber and Lyft can take care of things from now on. Imagine what happens next: Everyone who rides the LA Metro Bus system suddenly crowds onto the 405 in an Uber, every passenger on New York’s L train has to hail a ride over the Williamsburg Bridge. The result would be total gridlock.

Uber and Lyft have some advantages in certain contexts. But car services can’t overcome urban geometry.

2. Even lightly-used transit beats heavily-used ride-hailing services

Not every bus is packed, but even a mostly-empty bus can use streets more efficiently than Ubercars. A bus carrying about 10 passengers per service hour is generally considered to be “low-performing,” TransitCenter points out. But that still beats the pants off ride-hail services.

“For an Uber or Lyft driver to serve ten people per hour,” writes TransitCenter, “it would mean the driver is picking up a new passenger every six minutes, physically impossible in American cities.”

3. Demand for transit peaks at different times than demand for taxis

If you look at when Uber and Lyft are most popular, it’s during the night, when transit runs less often. Meanwhile, transit is at its fullest during the a.m. and p.m. rush. Not many people use Uber and Lyft for regular commuting.

Transit and ride-hailing services can complement each other — especially at times or in places where transit is weaker. But don’t be taken in by anyone predicting the end of transit — buses and trains aren’t going anywhere.


Note that pics like the above, while they viscerally illustrate the point, actually understate the difference in required space… cars require much more space, because they need to be separated by a long distance at any speed. For instance, using the “3 second rule” [of thumb], a car going 30mph needs 40-50m of lane.  This is a key factor in the vastly greater space-efficiency of mass transit: the inter-vehicle space required for safety is amortized over a huge number of passengers.

  • Even more than that, actually: when those people arrive at their destinations, the people in the cars will need almost an acre of land for parking, and the bus riders… won’t.

Uber is going to go bankrupt. People can’t get around by taxi because it takes an expensive form of transportation — the automobile — and makes it even more expensive by adding a driver. That’s why they are so desperate to go driverless before going under. Uber is caught between high costs for riders and low wages for drivers. It is losing massive money.

However, the system Uber and Lyft created is socially valuable, if it could be altered for use for actual dynamic carpooling.

With much lower fares, drivers just trying to get enough money to cover the cost of the car while making trips they would have made anyway, a fee of perhaps 25 cents per ride, and a company with a much lower market cap. And that could be useful in places where population densities are too low for transit without deep subsidies.

Perhaps they’ll be bought out of bankruptcy for that purpose. But unless you can get a ride for, perhaps, 50 cents more than a transit ride, with a company that isn’t losing money on every ride, it isn’t going to work.

Of course if they do go driverless, and the subway still relies on workers who earn (including retirement benefits) multiples of what the riders do, wouldn’t that be ironic?

  • Google Waze was trying this in the Bay Area, paying drivers the IRS rate of 50 cents-ish a mile. The problem is you need a critical mass for this to work – definitely in the peak hours, not so much late at night or on weekends.

  • News on their financial mess here.


While I share the writer’s skepticism of taxis (may as well call them that) as a workable solution, it’s not for the reasons in the article. For #1, no one is seriously promoting taxis-as-transit is aiming at heavily-trafficked routes like the L-train. That’s clearly ridiculous. They’re aiming at lightly-used lines that require large subsidies.

#2 is closer to the mark, but you don’t necessarily need to replace buses 1-for-1 to make it worthwhile. Even if replacing a bus requires two or three vehicles (and I’m assuming here that they’ll be using “Uberpool” or the equivalent), the costs might still be less overall, since buses are expensive to run. This is a question of numbers. And the service provided will probably be much better than a fixed-route, low-frequency bus, which is worth something. Using the streets efficiently isn’t likely to be a concern on a low-usage route, and while a full bus will beat the pants off single-occupant vehicles, a bus with only a few passengers may be worse congestion-wise than 2-3 cars. Again, all this only applies to very low-ridership lines.

#3 seems like a non-sequitur. If transit agencies started subsidizing them for commuters, they would surely become a lot more popular for commuters. It’s not really relevant how popular they are now.

I know I’ve said this before, but that makes a case *for* Uber, not against it. If we remove private car ownership and go to a shared car model, we can repurpose so much parking for housing that we can increase density. That makes for shorter overall trips – trips that are much more likely to be by walking or bike! And even if people decide to take an Uber, the overall VMT is lower because the trip is shorter.

I do agree wholeheartedly for major trunk lines we should use buses and trains. Right now, a lot of those forms struggle because of the last mile problem. “How do you get to the train?” In San Francisco that was a cobbled together fabric of walk, bus, taxi, bike, skateboard, whatever. Walking worked if you were close. Bus worked if you were on the line that went to the train and was frequently time consuming and riddled with missed connections. Taxis are expensive. Bikes have theft issues, so we cobbled together random solutions of bike lockers, valets, and using up space on the train to run 80 bikes per train – which is a good thing for last mile on both ends, but for many the only reason people are taking the bike on the train is to prevent theft. Using Uber as a seamless last mile could revolutionize the way people think about taking these big trunk rapid lines.

I do like the skateboards though.

Good solutions to the last-mile problem include increased density around stations (making walking viable for more people) and Japanese/Dutch-style high-density bicycle parking at transit stations.

Bikes on transit:

Sure it’s practical. It just takes a bit of design work.

BART is configuring its new trains to have three doors, I can imagine some smart designer coming up with a train car whose entire wall rolls up or something similar.

And as for taking a lot more space, it’s three seats, one for the passenger and two for the bike, but really, if you pile them up, it’s about 1.5 for the bike.

You’re going to save at each end, where you end up not running local buses. it makes up for reduced capacity per car requiring more cars per train, or more frequent trains. Or interuban busses with fewer seats.

The problem is that generally, interurban transit is under a separate budget from the local systems that feed it, so the administrators don’t see that saving and are not motivated to make it happen.