Looks like Uber has set up a system where either the riders will be unhappy with high surges, or drivers will be unhappy with the normal pricing on big nights. I never thought abut that side of it.
I'm not too surprised. Uber has created a perception that their service will be unaffordably expensive at peak times like this, so people don't even check.
I actually did check, and they were running at a 3.5x surge price in NYC. I stood on a street corner for 30 seconds and flagged down a yellow cab easily, so you might say their offer was lacking.
EDIT: reading the article again, I think there's something more of interest in here. Drivers are annoyed at their lack of surge earnings, which makes me wonder to what extent they're unhappy with how much they make at regular prices, and how much they depend on surge pricing windows to really make money.
I'm not exactly sure how Uber's surge algorithms work, but this seems like it should be mitigated the smaller the price and time adjustments are made. It's not like they don't have the data to do that.
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.
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.
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)?
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.
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.
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.
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.
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.
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).
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?
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.
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.
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".
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.
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.
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.
A bidding system could conceivably address this, or just a more dynamic pricing system that reduced fare multipliers as demand sagged. (Or does it work this way already?)
Or, just fixed fares and sometimes you can't get a ride. Seems to work for the rest of the transportation system.
Yeah, the fare multipliers are already dynamic and NYE's riders got a good deal. The issue is that we've had some highly-noticed surge pricing stories so all of the customers figured "oh yeah, it's going to be horrible surge pricing again and I'm going to be stranded at the FOO Bar after they close." So people changed their plans to either include a DD, use public transit, use traditional taxis, or go somewhere within walking distance.
Me and my friends stayed in and drank board games and played tequila.
a more dynamic pricing system that reduced fare multipliers as demand sagged
This is supposedly exactly how it works. Fare multipliers are updated every couple of minutes and fluctuate according to number of available drivers on the road (supply) and number of ride requests (demand).
I am a good example for this. I was planning to use Lyft but then scared of surge pricing and drove my own car. I was lucky to find free street side parking but even paying $40 for parking wouldn't be that bad compared to Lyft/Uber rates that night.
Most people avoid taking their own cars because they're going to be getting wasted, so I'm not surprised you were able to find street parking. I'm guessing your New Years was a little less "festive" than others' :) (I hope it was!)
> "There should be some regulation on the numbers of drivers you allow to Lyft," another commenter wrote. "Otherwise it may seem that you are taking advantage of the fact that we are not paid salaries, and you reap the benefits regardless if it effects us."
It will self regulate. If there isn't enough money in it, there will be less drivers. When ridership goes up, it will attract more drivers.
The entire purpose of surge pricing is to ensure that there is always a care when you want one. The purpose is not profit maximization, but ensuring there will never be a time that I go to Uber, request a vehicle, and not have one appear within a reasonable period of time.
If Surge pricing is working then there should be a large number of empty cars waiting for rides. Those cars are incentivized to show up by getting paid more than they normally would for a ride of the same distance.
I checked the Uber app in my city at 12:30AM and the rate was 2.0x, closed the app, reopened it until it was at normal 1.0x around 12:50AM, 4 USD was my bill , I used it again at 3:30 AM and the surcharge was 1.5, used it anyway since I was really tired, my bill was 6 USD.
Well the assumption of supply adjustments to meet demand in real time that is given for Surge Pricing is just unrealistic. A driver must make a commitment (hours or days before) to be available to drive at a given time. Of course they have no idea what the surge multiplier is, so they'll have to guess. As a result the number of cars on the road is driven by the _expectation_ of the multiplier, rather than the actual multiplier.
This affects the demand side too: people remember stories of people getting burned by huge surge multipliers, and as a result adjust their behavior (being more willing to call a normal cab, driving instead of taking a cab). Here again the actual surge multipliers matter less than the expectations. By providing reasonable estimates, they could set surge multipliers that would ensure a high supply of drivers and demand for them, instead of having the curves moved out of whack by overestimation.
This is a feature, not a bug. :) Uber has created a market with dynamic pricing, but manages to keep both ends of the transaction in the dark until the last possible moment. This information asymmetry is precisely why they are making money hand over fist.
I don't understand why this information asymmetry is more important to their profits than the network effect of controlling a very widely used marketplace?
It's a feature, and why passengers didn't even bother.
It's a dangerous game. With the cost of ride essentially being an unknown, ranging from cheaper than a cab to many times the cost of a cab, I often wouldn't bother either. People prefer consistency in what they pay for an experience, and when it's inconsistent, I lose trust that the value will be consistent (ie. Uber for most people I believe is attractive if the experience is better or equal in cost to a cab, but terrible when the cost is significantly higher than the convenience/better experience.) At least with a cab I am (fairly) certain of the value.
It's a "feature". :P In the same way that it is a feature that the cab driver doesn't have a permit or the proper insurance. Is this really how the new blackmarket is going to work? Do something illegal nationwide with average citizens running it and you are good to go? Maybe I am just being negative. Sorry.
It's a feature that optimizes for short term profit at the expense of the long term: memory of last year served to reduce demand this year, and memory of this year will serve to reduce supply in the future. Part of me wonders if they didn't tweak the surge multipliers in a misguided effort to appease drivers' expectations.
Given that supply and quantity demanded seemed badly mismatched this seems like a failure of the pricing mechanism, one which would not occur in a free market
Uh... Did someone in the article seriously ask for Uber/Lyft to artificially limit the number of cars on the road? Isn't that one of the problems with traditional taxis? They never learn. Meet the new boss, same as the old boss.
I'm glad to be living in a city with a functional public transit system (NYC) so I no longer have to be forced to be robbed by Uber (SF).
I live in Brooklyn, so it's still a choice between a 20 minute drive or an hour and a half on public transportation.
Actually, I think the M60 SBS should be the actual solution to this, but I haven't ridden it yet. I fly out of JFK even if I have to connect somewhere. (That's 18 minutes on LIRR for me.)
Well that makes sense then. Brooklyn to central and northern Queens is one of the worst parts of the whole system. I live in Woodside and have friends in Park Slope, so I definitely feel your pain.
All of this has happened before, and all of this will happen again ;)
This is basically a retelling of the cab industry origin story. The point is that this is an industry where the barriers to entry are so low, and the supply side so abundant (with the collapse of low skill labor economies across the US, ever more so) that the equilibrium state means everyone goes hungry.
And more relevantly for us silver spoon Silicon Valley types, more and more corner cutting to eke out a profit.
It's interesting watching us re-learn the motivations behind some regulations. This is why medallions were created, a far cry from the mustache twirling evil shenanigans that we'd like to paint on regulations.
The medallion system emerged from real problems. It may not be a particularly good solution, but an alternative that does not address the problem at all seems equally misguided.
I feel like on NYE, in particular, most people don't go out without already having made a plan for how they're going to get home, so even if the actual price turned out to be lower than expected, that might not have boosted demand the way one might expect because many people had already made other transportation arrangements in anticipation of Uber being unaffordably expensive.
Flywheel offered a flat rate of $10 for all cab rides on New Year's Eve. By contrast, Uber sent a mass email warning riders that New Year's Eve would see a spike in passenger demand, potentially skyrocketing prices to over $100 for a ride after midnight.
If you want to have no customers, then yeah, this is how you would do that.
That article was pretty much only comprised of anecdotes about drivers being upset about something... They even directly say "hard numbers on tech riders were not available."
I want to see the actual data and how Uber, or Lyft came out after NYE. Wasn't the algorithm supposed to help stabilize the supply of Uber cars on the road? I doubt it is designed simply to gouge people when demand is high.
It kind of sounds to me like the media pumped up this idea of "OMG look at how screwed these few people were by surge prices!" Then Uber responded to the critics by issuing warnings about high surge prices on NYE. This further solidified this idea that the price would always shoot up on big nights no matter what. So now you have everyone who's ever driven for Uber, plus a bunch of newcomers all going out driving thinking they are all going to be cashing in on $200 fares. Supply goes up, surge goes down.
Or maybe surge doesn't go all the way down. Uber would have no reason to make the primary function of their algorithm to make drivers more money, or stabilize driver income. They are trying to increase market share and make money. Maybe they think the NYE demand is very inflexible and thus they can restrict the surge pricing to stay above a certain level and not have much of an effect on demand. This would, however, probably only exacerbate the initial problem with oversupply, as more drivers would see surge pricing in effect and start earlier, and stay out later.
There isn't even any anecdotal evidence in the article about the passenger side of the experience, they only have stuff from Lyft/Uber drivers.
People act like Uber/Lyft should be nice to the drivers and treat them well, when in reality they are clients, not prized employees. Uber's relationship with the drivers is a business relationship that is very carefully maintained and fine tuned. If you showed me a similar company that treated drivers very well and was still able to keep prices from exploding, I would gladly use that service over the current players, but it just seems to me that they are doing what makes sense given the current marketplace. Isn't it supposed to be a lot easier gig to be an Uber driver than a city cab driver?
59 comments
[ 0.21 ms ] story [ 42.8 ms ] threadI actually did check, and they were running at a 3.5x surge price in NYC. I stood on a street corner for 30 seconds and flagged down a yellow cab easily, so you might say their offer was lacking.
EDIT: reading the article again, I think there's something more of interest in here. Drivers are annoyed at their lack of surge earnings, which makes me wonder to what extent they're unhappy with how much they make at regular prices, and how much they depend on surge pricing windows to really make money.
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.
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.
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?
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.
Well actually they don't because they overrode the algorithm instead of letting it do its job.
The culture at Uber is actually "greed".
Well done Uber.
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).
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.
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.
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.
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.
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.
Or, just fixed fares and sometimes you can't get a ride. Seems to work for the rest of the transportation system.
Me and my friends stayed in and drank board games and played tequila.
This is supposedly exactly how it works. Fare multipliers are updated every couple of minutes and fluctuate according to number of available drivers on the road (supply) and number of ride requests (demand).
It will self regulate. If there isn't enough money in it, there will be less drivers. When ridership goes up, it will attract more drivers.
Isn't that the point?
If Surge pricing is working then there should be a large number of empty cars waiting for rides. Those cars are incentivized to show up by getting paid more than they normally would for a ride of the same distance.
This affects the demand side too: people remember stories of people getting burned by huge surge multipliers, and as a result adjust their behavior (being more willing to call a normal cab, driving instead of taking a cab). Here again the actual surge multipliers matter less than the expectations. By providing reasonable estimates, they could set surge multipliers that would ensure a high supply of drivers and demand for them, instead of having the curves moved out of whack by overestimation.
I think it's probably one of the 'things that could work better' but unlikely something the company makes much money from.
It's a dangerous game. With the cost of ride essentially being an unknown, ranging from cheaper than a cab to many times the cost of a cab, I often wouldn't bother either. People prefer consistency in what they pay for an experience, and when it's inconsistent, I lose trust that the value will be consistent (ie. Uber for most people I believe is attractive if the experience is better or equal in cost to a cab, but terrible when the cost is significantly higher than the convenience/better experience.) At least with a cab I am (fairly) certain of the value.
http://imgur.com/yG5Ny1Q
It's all speculative because how they implement surge is opaque.
I'm glad to be living in a city with a functional public transit system (NYC) so I no longer have to be forced to be robbed by Uber (SF).
Unless you want to go to LaGuardia airport.
Actually, I think the M60 SBS should be the actual solution to this, but I haven't ridden it yet. I fly out of JFK even if I have to connect somewhere. (That's 18 minutes on LIRR for me.)
This is basically a retelling of the cab industry origin story. The point is that this is an industry where the barriers to entry are so low, and the supply side so abundant (with the collapse of low skill labor economies across the US, ever more so) that the equilibrium state means everyone goes hungry.
And more relevantly for us silver spoon Silicon Valley types, more and more corner cutting to eke out a profit.
It's interesting watching us re-learn the motivations behind some regulations. This is why medallions were created, a far cry from the mustache twirling evil shenanigans that we'd like to paint on regulations.
The medallion system emerged from real problems. It may not be a particularly good solution, but an alternative that does not address the problem at all seems equally misguided.
If you want to have no customers, then yeah, this is how you would do that.
I want to see the actual data and how Uber, or Lyft came out after NYE. Wasn't the algorithm supposed to help stabilize the supply of Uber cars on the road? I doubt it is designed simply to gouge people when demand is high.
It kind of sounds to me like the media pumped up this idea of "OMG look at how screwed these few people were by surge prices!" Then Uber responded to the critics by issuing warnings about high surge prices on NYE. This further solidified this idea that the price would always shoot up on big nights no matter what. So now you have everyone who's ever driven for Uber, plus a bunch of newcomers all going out driving thinking they are all going to be cashing in on $200 fares. Supply goes up, surge goes down.
Or maybe surge doesn't go all the way down. Uber would have no reason to make the primary function of their algorithm to make drivers more money, or stabilize driver income. They are trying to increase market share and make money. Maybe they think the NYE demand is very inflexible and thus they can restrict the surge pricing to stay above a certain level and not have much of an effect on demand. This would, however, probably only exacerbate the initial problem with oversupply, as more drivers would see surge pricing in effect and start earlier, and stay out later.
There isn't even any anecdotal evidence in the article about the passenger side of the experience, they only have stuff from Lyft/Uber drivers.
People act like Uber/Lyft should be nice to the drivers and treat them well, when in reality they are clients, not prized employees. Uber's relationship with the drivers is a business relationship that is very carefully maintained and fine tuned. If you showed me a similar company that treated drivers very well and was still able to keep prices from exploding, I would gladly use that service over the current players, but it just seems to me that they are doing what makes sense given the current marketplace. Isn't it supposed to be a lot easier gig to be an Uber driver than a city cab driver?