3 ms·

I saw something strange here in Seattle. Usually there is surge pricing whenever there are few cars around. However I checked multiple times on New Years Day at

I saw something strange here in Seattle. Usually there is surge pricing whenever there are few cars around. However I checked multiple times on New Years Day at around 12:30AM and the surge pricing was >4.5x, but there were loads of UberXs around. Like, 10 cars within a couple blocks. And if I dragged the gps marker 10 streets over, another huge cluster of cars would show up. It looked to me like Uber overrode the algorithm and jacked up the prices really high, even though there wasn't such high demand for Ubers. I regularly take Uber at 9AM and there will often be no cars in the area (wait time of 8 minutes when usually it's <2 minutes) and the surge shows only 2.0x.

skuhn 12y ago
Uber claims that New Years Eve is their highest traffic day. I could imagine they might override the surge algorithm (or give it a nudge) to provide a more consistent price based on their historical data. I've encountered a number of situations where Uber's surge prices annoyed people not simply by how expensive it was, but because the person standing next to them got 2.5x and if they had waited another 10 minutes they might get 1.25x or 8x. The non-repeatability can be frustrating and feel unfair. However, just because NYE 2014 naturally surged to 4.5x doesn't mean that NYE 2015 will, because things change. Not the least of which is the (small, but vocal) backlash against Uber that this article is a part of.
- joezydeco 12y ago
  Uber claims that New Years Eve is their highest traffic day But that was in previous years, right? A lot of things have happened in the last 12 months to change this NYE's outcome: more competition, more of their own cars on the street, a year of bad publicity against Uber, and then telling everyone that there will be surge pricing in effect (instead of letting it happen algorithmically)? How is anyone surprised by the outcome?
  - untog 12y ago
    It probably speaks (at least partially) to the culture at Uber - that data and algorithms rule all. For the most part that's correct, but things often get unpredictable when you involve humans - no matter how fair surge pricing is when looking at the data, if people feel like it is unfair, they'll object. Right now I don't think Uber does a good job at, for want of a less fluffy term, "feeling sentiment". They're extremely good at parsing past data and extracting conclusions, but that's about it.
    - kenzokai 12y ago
      > that data and algorithms rule all Well actually they don't because they overrode the algorithm instead of letting it do its job. The culture at Uber is actually "greed".
      - untog 12y ago
        Well, that's what we think they did. We don't know that. But even still, if it were true it does just highlight that when they're not relying on existing data they really don't have good instincts.
    - krapp 12y ago
      If people felt Uber's price was too high, regardless of the calculus, I'd put the blame for that on Uber. People had other more affordable options available and they used them, rather than deal with Uber's surge rate. That's not unpredictable, that actually quite rational.
meritt 12y ago
Yup, sounds like you're right. What should have been an automated algorithm based on supply & demand was instead hand-tuned by an idiot and fucked everyone over. Well done Uber.
- potatolicious 12y ago
  This is why I like Lyft a lot more. They don't take a cut of the surge portion of the fare - and removes the incentive for them to override the surge multiplier for profits' sake.
  - adamio 12y ago
    This used to be the case, but Lyft now takes 20% of the total fare, including surge
    - potatolicious 12y ago
      Aww crap. Guess I stick to regular taxis from now on...
- prostoalex 12y ago
  I'd argue that it's a special night, one of the few nights where people like to make plans in advance. Therefore traditional push notification to get behind the wheel and start driving wouldn't work if the driver already made plans to celebrate New Year's with the family (and perhaps consume alcohol).
jfoster 12y ago
Lots of cars within a small area is a snapshot of one point in time, though. What you don't get to see (but presumably Uber's algorithm does) is the liquidity; how quickly are cars being taken, and how quickly are new cars becoming available?
applecore 12y ago
The problem is that customers and drivers don't actually set the price. The market would operate much more efficiently as a double auction: potential passengers submit their bids and drivers simultaneously submit their ask prices, and Uber continuously chooses some price that clears the market.
- seanmcdirmid 12y ago
  This is how didi dache works: you set the tip you will provide, taxis won't bother coming during rush hour for just a 5 kuai tip, but for 10 kuai I don't have a problem (on top of a 40 kuai base fair). These are all official taxis however and the gov doesn't like the tip function since it defeats the purpose of fixed rates for obvious reasons; it is also almost impossible to get a taxi during rush hour without the app now. But the drivers love it and I doubt it's going away. 10 kuai is less than 2 bucks and they don't make enough money anyways.
  - twic 12y ago
    Interesting. As a would-be passenger, is there any way to tell how much you'd have to tip to get a ride? Or to get a ride within a given period of time? It'd be very useful to see that curve - "well, if I tip seven kuai, I'll get a ride in ten minutes, but if I tip nine kuai, I'll get it in five minutes, which sounds good, but to get a ride in three minutes I'd have to tip twenty-five kuai, so forget that".
    - seanmcdirmid 12y ago
      You specify the tip in the app at time of your hail, so the drivers know what you are willing to pay on top of the meter (my WP app provides 5, 10, and 20 kuai tip options; the iOS app is more flexible). The taxis are responsible to replying to your hail, and you are given 120 seconds per request, so if no one replies, your request simply fails. You are supposed to hail when you need the taxi, so there is no time/price tradeoff to be made. There is another service (didi chuanche?) that is higher end and more expensive (they come in nice rental cars) that you can reserve ahead of time, but the government has made it basically illegal (its still going, but they have to be smart about where they drop you off). Note that tipping is not a part of Chinese culture, you aren't usually expected to tip. This is more like an incentive for them to come pick you up. I was already tipping 10 kuai anyways before I started using the app, it just felt like the right thing to do.
prezjordan 12y ago
And the main argument is: there may very well have been loads of cars around, but people still paid $x for them, so that's what they cost! Supply and demand. To me it's a little deceiving, because Uber frames it like this grand algorithm determining the pricing, when in reality I feel there's a substantial human element in it.
- skuhn 12y ago
  From the outside looking in, it's hard to say how much of a human element there is. Certainly it must operate algorithmically (hands-off) most of the time, because Uber operates in way too many markets for someone to babysit each one (including significant sub-regions of each, such as Manhattan and SOMA) around the clock. My theory is that they may have goosed the algorithm based on prior years, or perhaps their algorithm considers historical data for all (or just important) dates. Creating an algorithm that correctly models demand / pricing for all regions is pretty tough, and there are always going to be situations where what seemed reasonable gives you incorrect results.

skuhn 12y ago

Uber claims that New Years Eve is their highest traffic day. I could imagine they might override the surge algorithm (or give it a nudge) to provide a more consistent price based on their historical data. I've encountered a number of situations where Uber's surge prices annoyed people not simply by how expensive it was, but because the person standing next to them got 2.5x and if they had waited another 10 minutes they might get 1.25x or 8x. The non-repeatability can be frustrating and feel unfair. However, just because NYE 2014 naturally surged to 4.5x doesn't mean that NYE 2015 will, because things change. Not the least of which is the (small, but vocal) backlash against Uber that this article is a part of.

joezydeco 12y ago
Uber claims that New Years Eve is their highest traffic day But that was in previous years, right? A lot of things have happened in the last 12 months to change this NYE's outcome: more competition, more of their own cars on the street, a year of bad publicity against Uber, and then telling everyone that there will be surge pricing in effect (instead of letting it happen algorithmically)? How is anyone surprised by the outcome?
- untog 12y ago
  It probably speaks (at least partially) to the culture at Uber - that data and algorithms rule all. For the most part that's correct, but things often get unpredictable when you involve humans - no matter how fair surge pricing is when looking at the data, if people feel like it is unfair, they'll object. Right now I don't think Uber does a good job at, for want of a less fluffy term, "feeling sentiment". They're extremely good at parsing past data and extracting conclusions, but that's about it.
  - kenzokai 12y ago
    > that data and algorithms rule all Well actually they don't because they overrode the algorithm instead of letting it do its job. The culture at Uber is actually "greed".
    - untog 12y ago
      Well, that's what we think they did. We don't know that. But even still, if it were true it does just highlight that when they're not relying on existing data they really don't have good instincts.
  - krapp 12y ago
    If people felt Uber's price was too high, regardless of the calculus, I'd put the blame for that on Uber. People had other more affordable options available and they used them, rather than deal with Uber's surge rate. That's not unpredictable, that actually quite rational.

meritt 12y ago

Yup, sounds like you're right. What should have been an automated algorithm based on supply & demand was instead hand-tuned by an idiot and fucked everyone over. Well done Uber.

potatolicious 12y ago
This is why I like Lyft a lot more. They don't take a cut of the surge portion of the fare - and removes the incentive for them to override the surge multiplier for profits' sake.
- adamio 12y ago
  This used to be the case, but Lyft now takes 20% of the total fare, including surge
  - potatolicious 12y ago
    Aww crap. Guess I stick to regular taxis from now on...
prostoalex 12y ago
I'd argue that it's a special night, one of the few nights where people like to make plans in advance. Therefore traditional push notification to get behind the wheel and start driving wouldn't work if the driver already made plans to celebrate New Year's with the family (and perhaps consume alcohol).

jfoster 12y ago

Lots of cars within a small area is a snapshot of one point in time, though. What you don't get to see (but presumably Uber's algorithm does) is the liquidity; how quickly are cars being taken, and how quickly are new cars becoming available?

applecore 12y ago

The problem is that customers and drivers don't actually set the price. The market would operate much more efficiently as a double auction: potential passengers submit their bids and drivers simultaneously submit their ask prices, and Uber continuously chooses some price that clears the market.

seanmcdirmid 12y ago
This is how didi dache works: you set the tip you will provide, taxis won't bother coming during rush hour for just a 5 kuai tip, but for 10 kuai I don't have a problem (on top of a 40 kuai base fair). These are all official taxis however and the gov doesn't like the tip function since it defeats the purpose of fixed rates for obvious reasons; it is also almost impossible to get a taxi during rush hour without the app now. But the drivers love it and I doubt it's going away. 10 kuai is less than 2 bucks and they don't make enough money anyways.
- twic 12y ago
  Interesting. As a would-be passenger, is there any way to tell how much you'd have to tip to get a ride? Or to get a ride within a given period of time? It'd be very useful to see that curve - "well, if I tip seven kuai, I'll get a ride in ten minutes, but if I tip nine kuai, I'll get it in five minutes, which sounds good, but to get a ride in three minutes I'd have to tip twenty-five kuai, so forget that".
  - seanmcdirmid 12y ago
    You specify the tip in the app at time of your hail, so the drivers know what you are willing to pay on top of the meter (my WP app provides 5, 10, and 20 kuai tip options; the iOS app is more flexible). The taxis are responsible to replying to your hail, and you are given 120 seconds per request, so if no one replies, your request simply fails. You are supposed to hail when you need the taxi, so there is no time/price tradeoff to be made. There is another service (didi chuanche?) that is higher end and more expensive (they come in nice rental cars) that you can reserve ahead of time, but the government has made it basically illegal (its still going, but they have to be smart about where they drop you off). Note that tipping is not a part of Chinese culture, you aren't usually expected to tip. This is more like an incentive for them to come pick you up. I was already tipping 10 kuai anyways before I started using the app, it just felt like the right thing to do.

prezjordan 12y ago

And the main argument is: there may very well have been loads of cars around, but people still paid $x for them, so that's what they cost! Supply and demand. To me it's a little deceiving, because Uber frames it like this grand algorithm determining the pricing, when in reality I feel there's a substantial human element in it.

skuhn 12y ago
From the outside looking in, it's hard to say how much of a human element there is. Certainly it must operate algorithmically (hands-off) most of the time, because Uber operates in way too many markets for someone to babysit each one (including significant sub-regions of each, such as Manhattan and SOMA) around the clock. My theory is that they may have goosed the algorithm based on prior years, or perhaps their algorithm considers historical data for all (or just important) dates. Creating an algorithm that correctly models demand / pricing for all regions is pretty tough, and there are always going to be situations where what seemed reasonable gives you incorrect results.