Three Lyft Rides in a Rare Snowstorm, and Musings on “Rate Your Driver”

Portland's transit mall five days after the snow fell.

Portland’s transit mall five days after the snow fell.

Portland’s epic snowstorm continues.  Five days after we got about 9 inches (30 cm) of snow, most streets are still coated with ice.  Again, this shouldn’t be judged by the standards of snowier cities; this kind of extended storm happens less than once a decade here.

In my last post I talked about my transit experience on the first day, when snow was still coming down hard.  Later that day, and twice the next day, I used Lyft– but since most Portland drivers work for both Uber and Lyft, I’m guessing an Uber experience would have been identical.

My three Lyft rides gave me five interesting data points:

  • A trip to my mother’s house in Portland’s hilly west side.  The driver showed up in a tiny car without snow tires or chains, and said he can’t climb hills.  To prevent discrimination against customers based on their destination, the companies don’t tell drivers the destination until after they accept the ride, so this guy was unable to say no to a destination that he couldn’t physically reach.  Likewise, I was unable to specify, when requesting a ride, that the driver needs chains or snow tires and probably 4-wheel-drive to get there.  That’s obviously a design flaw in Lyft’s systems.
  • A trip back from the same hilly area.  The driver showed up in a small car with no left rear view mirror.  He told me another car knocked the mirror off and drove away the previous night.  He plans to get it fixed, he said, but meanwhile he has to keep driving, so he just glances over his left shoulder now and then.
  • On the same ride: The driver told me he lives in Los Angeles but is in Portland for a few months for some reason.  He’d never driven in snow or ice before, and clearly didn’t understand the risks.  I had to explain that in these conditions, you allow even more room behind the car in front of you.  “But people will cut in!” he said.  “Yes,” I said, “but you won’t die.”
  • … but on an earlier trip back from the airport, the driver was a very recent immigrant from Ghana who had also never driven in snow before.  I was surprised to learn this, because his snow driving skills seemed perfect — which maybe just means he was quite properly terrified and being very careful.

How am I to “rate your driver” in some these cases? The Ghanaian guy gets 5 stars of course, but what of the other two?

Should I give a low grade to the a driver whose car is unsuited to these conditions, at least in the hilly part of town?  Obviously this is mostly a policy failure, but should I declare, from my perch of authority, that he should have had chains?

What about the driver who didn’t have a left rear view mirror?  Well, that’s plainly illegal and unsafe, but as I talked with him, I wondered if he could afford to not drive until he can get it fixed.  As it was, he was also the tailgater, so it was easy to rate him low, but what if he’d been a great driver otherwise?

As I think about this, and about how the “rate your driver” scores are processed, I realize that I pretty much have to give 5 stars for anything other than obvious rudeness (unexplainable by cultural difference) or reckless driving.  That’s because I can sense how much pressure these people are under, how few options they have, and how devastating even a 4-star rating can be.  For all I know, someone’s kids are going to starve if I tell a guy to quit driving until his mirror is fixed.  On the other hand, for all I know, he’s doing fine and is just risking his life and that of his passengers because he’s greedy.

Once you open that window into considering the real causes of problems, and the real impacts of ratings, it’s hard to close it.

This is an instance of a more general problem with all of the “how did we do?” surveys that fill my inbox every day.  They really want my opinion of front-line staff, but often I can see that my negative experience was a matter of management.  The obvious example is restaurant or hotel staff who are harried and unresponsive because the management has decided to have too few people on duty.  I’ve learned to be careful about this.  Unless I’m sure that the frontline person was entirely responsible for the outcome, the worst situations get five stars, plus maybe a little note that nobody will read.


14 Responses to Three Lyft Rides in a Rare Snowstorm, and Musings on “Rate Your Driver”

  1. Isaac Rabinovitch January 16, 2017 at 6:50 pm #

    > “But people will cut in!” he says. “Yes,” I say, “but you won’t die.”

    I believe that logic applies even when conditions are perfect.

  2. Jason January 16, 2017 at 6:58 pm #

    It is interesting that these sharing economy services go with “rate your driver” (or homestay, etc) rather than “rate our service”.

    I suppose they can watch the average of driver ratings wax and wane as they alter their terms of service and infer when and where policy changes are improving ratings. But it would be confounded: Much of the time, the driver is just neutral, does nothing wrong, and gets 5. Sometimes, policy quirks make the driver’s behaviour seem relevant and they get a low rating. If the service boots out this second group, they may never notice that it was their policy at fault.

    Furthermore, the knowledge that they kick out drivers with low scores confounds ratings. We rate highly those we like and those we pity.

    They might imrpove the rating system if they communicated better what should be included as “driver-dependent” and what is system-dependent, and also what they do with the rating.

    It would be better if the communicated how they intend to use th

  3. calwatch January 16, 2017 at 9:17 pm #

    Rating systems are more pernicious where there is no direct supervision of the person. For McDonald’s and Starbucks surveys I regularly rate service “good” not excellent, and on Starbucks surveys especially when they have that “barista attempted to make a personal connection with me” I routinely rate it negative. The same goes for those hotel surveys. But I assume that there is a shift manager or store manager who sees these people more frequently than I do. For ride services, there is zero supervision at all and the passengers have to be the supervisor.

  4. asdf2 January 16, 2017 at 9:57 pm #

    I definitely feel the pressure to give 5 stars unless there’s a specific reason not to. Some examples:

    1) Smelled cigarette smoke in the car, needed to open the window to make the trip bearable. Rating: 3 stars
    2) Driver made a wrong turn, adding several minutes to the trip. Rating: 4 stars
    3) Driver missed freeway exit, but apologized and ended the trip immediately so I wasn’t charged for the

    additional miles from getting off one exit too late. (Trip ended up being slightly cheaper than normal). Rating: 5 stars, since driver was already getting less money than normal for the trip.

    4) Lyft Line had the wonderful idea to match me up with a rider traveling in the opposite direction. Rating: 5 stars, since this was the software’s fault, not the driver’s fault.

    (Note: This is a fundamental problem with so-called “flexible” transportation systems; unlike a normal transit bus, which follows a fixed route and attempts to keep a fixed schedule, with a “flexible” system, it is a complete crapshoot how long your trip is going to take, even on a Sunday morning with almost no traffic. In this case, it was especially weird, since the “ride-matching” algorithm picked a ridematch that saved essentially zero vehicle miles traveled compared to the two of us traveling in separate cars. It might have made some sense if there were literally no other drivers around to pick up the other person, but it was a relatively slow weekend morning, and no surge pricing was in effect at the time, so I find that explanation rather unlikely).

  5. Dexter Wong January 16, 2017 at 10:08 pm #

    Yup, you have a choice of giving a rating that will give a poor unprepared driver a dismissal or saying everything was OK when it was not (a poor choice, indeed).

  6. Kakurady January 17, 2017 at 6:33 am #

    YouTube used to have a 5-star rating system too, but most people only used 1-star and 5-star, just like Uber. Which is why they switched to a thumbs up/down system. Uber seems to have tested it.

    One other interesting way that Uber is trying to solve the problem of 5-star “feedback” system not conveying enough information (and that freeform comments are hard to write on mobile): “Expert Navigation”, “Great Conversation”, “Awesome Music” etc.

    There’s no “Responsible Driver” or “driver is unequipped for the ride” though. And who would give a public “compliment” like “faulty equipment”, if it would mean the driver won’t be able to bring home dinner tonight?

  7. Ted January 17, 2017 at 12:26 pm #

    I can’t find Oregon law that requires EITHER outside mirror. Strange.

  8. Ari Ofsevit January 17, 2017 at 3:05 pm #

    If there were sensible regulation at the state level, you’d think that states (or even metro areas) would require in-state vehicle registrations to operate a TNC vehicle. This would have several benefits:

    1. Local knowledge of roadways and conditions. As Jarrett points out, there’s a driver up from LA in Portland for a few weeks/months. Which is fine, except that there is no barrier to entry for someone coming to a new city and driving with no local knowledge of roadways and conditions. Especially in busier cities with more convoluted street and route patterns, a talking GPS unit only gets you so far.

    2. Taxes and fees: why should vehicles operated in state x be registered in another state while the wear and tear takes place in another state. This is especially interesting somewhere like Boston, where there a number of cars registered in Rhode Island and New Hampshire plying the streets. Harder in areas that straddle state borders, though. (Taxicab medallions based on city boundaries are likely over-regulation, but a local registration doesn’t seem like a particularly high bar.)

    3. Emissions: many vehicles in non-coastal states don’t require annual emissions tests. What’s to keep someone with a Nebraska registration from driving to San Francisco with a polluting car and driving Uber for a while, keeping the vehicle registered at home? Sure, it’s not exactly legal, but this wouldn’t be hard to regulate.

    Common sense. I won’t hold my breath.

  9. Theo January 17, 2017 at 4:22 pm #

    Don’t forget the drivers also rate you too, as a passenger. So I presume that Lyft keeps a running total of how (racist / troublesome / mean) each passenger is and responds to ratings accordingly. I actually think this check-and-balance system works quite well.

  10. el_slapper January 18, 2017 at 2:28 am #

    Most rating systems are biased. We had one in my firm recently, and it was impossible to make sytemic remakrs(i.e. we need unified ergonomics on our product). Instead, the way questions were asked sounded like “who is to blame”.

    This Lyft rating is exactly the same, as it aims to target at specific problems, not systemic ones(the ones that would need real hard work from the brass)

  11. Federico January 20, 2017 at 7:26 am #

    Ghana is a rainy region similar to my own city, so the Ghana guy probably was applying the same cautions he have in a 100mm/hr rainstorm where you can’t see 30m forward

    • Murloc January 25, 2017 at 5:12 am #

      yeah aquaplaning is no joke, if you lose traction you need to have that space to let your car do its own thing and recover even if the guy before you slows down.

  12. Mike January 26, 2017 at 4:37 am #

    Doesn’t anyone use normal cabs anymore – where you don’t have to do ratings?

  13. Bobs Uncle November 30, 2018 at 11:41 am #

    Both Uber and Lyft consider any rating below 5 to be unsatisfactory. 5 = satisfactory 4 = unsatisfied 3 = not paired with again 2 = flag threshold 1 = flag.