Uber’s Matching Algorithm Optimizes Corporate Profits, Not Driver Satisfaction?

What People Hate About Being Managed by Algorithms, According to a Study of Uber Drivers

Uber, Lyft, and drivers for other TNCs (transportation network company) do not have a traditional boss… and that’s a good thing… I think?

The article referenced for this blog suggests not having a direct-report human boss is perceived by some drivers as more negative than positive… no I’m not pulling your leg! They’d rather let a human manage their rideshare work instead of relying on the Uber matching algorithm.

This quote from the article suggests the researchers did more than a little due diligence collecting data:

“We collected data by informally and formally interviewing 34 drivers, observing drivers in action, analyzing more than 1,000 online forum posts, and reviewing media coverage of Uber in several waves between December 2015 and September 2018.”

Interviewing only 34 drivers seems a little slim to me but let’s dig in and see what the article’s “three areas of consistent complaints” may tell us about being “managed” by software rather than a human supervisor.

Uber Drivers Dislike the Algorithms Because...

The first of the three is:

[1] Constant Surveillance

From the article:

“…the app tracks their [drivers] GPS location, speed, and acceptance rate of customer requests. It instructs them which riders to pick up where, and how to get to the riders’ destinations.”

Uhhh… okay… but how else could the basic function of a ridesharing software application work?

Before becoming a Lyft and Uber driver (in April 2016) I always kept the GPS function of my smart phones turned off unless there was good reason to turn it on… like using one of my phone’s installed navigation programs to get somewhere. 

I always say “No” when an application asks permission to know my location unless there is good reason to say “Yes.” 

When I download a restaurant’s application in order to earn points toward future discounts, I do not accept the premise the application needs to know where I am… I only put the application on my phone because few chain restaurants use frequent customer punch cards anymore… my goal is to earn points and I don’t trust the applications on my phone won’t track me and use/share my locations for other purposes.

In the past 3+ years as a rideshare driver this is still true for me, I don’t share the ability to track my location freely, but I accept the reality that Uber and Lyft driver applications won’t work if they don’t know where I am.

I don’t turn my phone’s GPS off anymore… like when I stop driving… just lazy I guess? 

Plus, my phone is rarely short on charge (GPS function is a power hog) because it spends most of it’s time plugged in sitting on my car’s phone mount.

The referenced article is not done on the “Constant Surveillance” point by simply pointing out that GPS function must be turned on when rideshare driving:

“If the drivers diverge from the app’s instructions they can be penalized or even banned from the platform.”

It’s common for me to see online posts from active drivers claiming they are being penalized for one thing or another, as example, for not accepting a sufficient percentage of passenger trip requests BUT I’ve never seen measurable proof what these drivers are saying is true.

Influencing the Uber Matching Algorithm

I’m willing to believe the Uber matching algorithms will somehow prioritize drivers who accept almost every trip request… but it’s not logical to believe the driver application software will, as example, make a passenger wait longer to be picked up because of software coding that is currently penalizing a driver?

Of course, I have reason to believe drivers who accept almost every trip will be favored by the algorithms because I accept almost every passenger trip request. 

I don’t understand drivers who say they pass on trip requests for any reason accept the pickup location being extremely far away… if I get a request that requires me to drive over 15 minutes to the pickup location I’m going to think about it for a second before accepting the request.

The way I see it if I let a trip request go by without accepting, I could be waiting without a paying passenger for… who knows how long?  This is especially true when I’m farther away from densely populated parts of where I drive in Denver, Colorado. 

When I’m on the outskirts of the city or just getting started for the day a 15-minute drive to the pickup location sounds like more money to me… so I accept and off I go. 

If I’m currently located in a densely populated part of Denver and/or driving at a time likely to have a lot of passenger trip requests I may pass on the long drive to my next pickup.

The “banned” from the driver application part is pretty much impossible for me to believe; however, once over 2 years ago one of the driver applications I use was misbehaving by not responding to my attempts to accept trip requests.  After 3 or 4 requests went by I was put in a “timeout” by the application, it logged me out and said I couldn’t log in again for 10 minutes.  

This “timeout” happened once and has never happened again. 

I’ve recently seen a “timeout” mentioned in driver forums connected to a driver making a conscious choice to not accept multiple passenger trip requests in a row so I suspect the “timeout” is still a “feature” in the driver application… but it doesn’t apply to me because again I accept almost every trip request.

Reportedly some drivers, drivers who have the ability to see the general direction and duration of a passenger trip request (Uber Pro Program), let short duration trips go by without accepting because drivers don’t earn much income from short trips.  In Denver a short (Minimum Fare) trip earns me about $3.50.  That would be $3.50 more than I had before I completed that trip.

If a driver did this multiple times in a row then conceivable they are put in a “timeout” by the Uber matching algorithms of the driver application.

Not accepting short trips is a driver behavior to which I don’t agree. 

The very nature of being a rideshare driver means short, medium, and long trips.  I’m not willing to sit around not earning anything waiting for a longer, more profitable trip request.

As a rideshare driver I never know when the next trip request will present… because there is no way to accurately predict the exact place and time a passenger will request a ride. 

Sure, when an event with lots of people ends there will probably be a lot of trip requests from that location, but I’m talking about when I’m in a random part of the city at a random time…

I accept almost every trip request simply because I’m out in my car to earn income… more trips means more income… more is better than less?

Denver is a good place to be a rideshare driver, it’s rare I’m waiting more than a few minutes for my next trip request but I’m not taking chances… accepting that short trip might put me in the perfect time/location to get a longer trip request… one I probably would not have received if I’d stayed in the same place letting short trips go by without accepting?

My point is as a driver there’s no way to know for certain there will be another, more profitable trip request coming soon so no way to know if the risk of passing on trip requests will mean the reward of earning more income during a rideshare driving “shift.”

The point of this blog is thinking about what it means to be “managed” by software not a human boss… for this blog I’m simplifying the specific methods I use every day to maximize my income and reduce my stress.  To get more details check out my book: How To Be A Lyft and Uber Driver – The Unofficial Driver’s Manual

Back to the referenced article, specifically the idea drivers could be:

“…penalized or even banned from the platform”

because the driver:

“…diverge[s] from the app’s instructions…”

Outside of something like the “timeout” I explained previously… any driver saying they were “penalized” by the driver application is presenting their best guess why [something] happened. 

Based on thinking through the logic behind the way the driver applications probably work, online posts from active drivers I’ve seen certainly sound like they were guessing… maybe they didn’t get any trip requests because there weren’t any passenger requesting trips in the driver’s specific location/time?

There are only two verifiable reasons a driver could be “banned from the platform.” 

ONE: A report from a passenger of driver behavior deemed inappropriate as defined in Lyft/Uber’s Community Guidelines

Uber Community Guidelines

Lyft Community Guidelines

TWO: When a driver’s average star rating falls below a certain level… neither Lyft or Uber publish a specific number but popularly believed to be around an average 4.6 star driver rating.

Drivers are not automatically disconnected by software coding in the driver application the nano-second their average star rating reaches the dreaded 4.6 stars… they will be contacted by the Lyft/Uber support organization who will work with the driver to improve their future star ratings from passengers. 

Of course, I have no evidence to prove it… but I believe these drivers are essentially put on a “program” (similar to traditional jobs when an employee needs to improve their work behaviors) and their future passenger star ratings and passenger comments are monitored by real, live human beings. 

If a driver is temporarily disconnected or even “banned” from the driver application, it is a choice made by humans not by the Uber matching algorithms.

I don’t want to spend a lot of time in this blog explaining how difficult it would be for an active driver to have an average star rating approaching 4.6 stars, but the referenced article brought it up:

“From our study we found that drivers found performance evaluations in the form of customer [star] ratings particularly frustrating. We believe this may be because they amplify Uber drivers’ negative feelings about constant surveillance with an additional layer of technology-mediated attention.”

To save us both time… consider this summary of how driver star ratings work:

  • Uber averages a driver’s most recent 500 trips where a passenger bothered to rate them
  • Lyft averages a driver’s most recent 100 rated trips

An Uber driver with 450 five-star ratings and 50 one-star ratings has an average 4.6 star rating.

A Lyft driver with 90 five-star ratings and 10 one-star ratings has an average 4.6 star rating.

Obviously, a driver is not going to receive only five-star and one-star ratings… the point is in order to reach the dreaded 4.6 star average rating requires a large percentage of low ratings from passengers.

I cover driver star ratings in detail in: How To Be A Lyft and Uber Driver – The Unofficial Driver’s Manual.

Drivers Don't Know How the Uber Algorithm Works

I see driver’s online appearing to have a lot of stress about, as example, a 4.8[whatever] star rating… in my book I explain why a driver’s star rating is more or less irrelevant… for this blog let’s move on to the referenced article’s point #2.

[2] Little Transparency

From the article:

“While the app is learning a lot about them, Uber drivers find it frustrating how little they know about the app.  They find the lack of transparency of the underlying logic of the complex algorithms frustrating, believing it to be an unfair system which manipulates them subtly without their knowledge or consent.”

I suspect my background working almost two decades in corporate America, specifically in technology companies, gives me a different take on this point. 

My corporate America job roles were such that I knew well “how the sausage was made” in an environment where most of the company’s workers only had deep knowledge in their specialized job roles.

As a rideshare driver I’m focused on what’s important to me in my specialized role as a driver and I’m unwilling to fault Lyft or Uber for doing what they think has the best chance to make them successful and I don’t agree with drivers posting online as if they should be the most important piece of the rideshare industry puzzle.

Without a doubt in the past three and a half years I’ve seen countless examples where a driver application appeared to be attempting to motivate me to, as example, to work more hours/complete more passenger trips.

Please note: I use the word “motivate” not “manipulate” because assuming the goal of a rideshare company and it’s driver application is to “manipulate” drivers is a value judgement (and a negative one at that) not a dispassionate assessment of the probable goal of the driver application’s software algorithms.

In other words, I choose to believe the rideshare companies understand their product cannot be delivered without drivers, so it makes no sense to purposefully take advantage of drivers which would result in driver frustration and probably less of what the rideshare companies want drivers to do… complete passenger trips. 

I heard once that in the absence of information it is human nature to believe a negative rather than a positive… thinking positively is for almost everyone (including me) a conscious choice.

Other aspects of my past that influence my writing and work as a Lyft and Uber driver… I have a computer technology focused college degree and I literally grew up with computers in my home. 

My father was a mainframe computer software developer and a personal computer hobbyist.  In our home growing up we had some of the earliest personal computers and I loved “farting around” figuring out what they could do and figuring out how they worked. 

My point is this: I believe I am able to detect what’s probably going on “behind the scenes” in the Uber matching algorithms so am better able to detect times when software may be trying to “motivate” me to behave in certain ways.

As a rideshare driver I’ve learned to make my own choices and to this point from the referenced article focus on what’s most important to me… earning income which is accomplished by completing passenger trips!!!

Rideshare drivers have lots of time to think (and I’ve always been someone who thinks a lot) so I completely understand how drivers could be frustrated because the way the driver applications work is not always obvious.

Best advice I can give… focus on the core nature of the job… completing passenger trips.

If my job at a restaurant was working the grill I’m probably going to be thinking about, as example, every aspect of the process from baby cow to piece of meat… but I accept that aspect of my nature as my personal “affliction…”

As a rideshare driver trying to understand how the Uber matching algorithms work to deliver passenger trip requests to a given driver is unlikely to result in receiving more or better trip requests. 

And more importantly Lyft and Uber are never going to share that kind of information so anything we think we figured out is really only going to be our best guess… so it makes no sense to stress about it?

Uber Algorithms Dehumanize Drivers

Let’s cover the article’s third of three “areas of consistent complaints” from their research:

[3] Dehumanization

“Drivers at Uber report feeling equally lonely, isolated, and dehumanized. They don’t have colleagues to socialize with or a team or community to be part of. They lack the opportunity to build a personal relationship with a supervisor.”

At the risk of sounding like I am making light of this final point… rideshare driving is a choice… if you want to work with a team or build a personal relationship with a supervisor then maybe rideshare driving is not the best choice for you.

Also true, most rideshare drivers only work part-time hours and have other sources of income and other work-related opportunities to feel less lonely and isolated.

This point from the referenced article has me scratching my head a little…

I’m a little confused why a lot of drivers might feel this way because as a rideshare driver I’m not isolated… I’m interacting with my passengers all day long.  I have tons of great conversations with all kinds of people every day I drive. 

Sure, it’s unlikely I’ll see a given passenger again, but personally I don’t see this as a negative.

I’ve had “bad” supervisors and worked many places where one or more employee was less than fun to work with and had to make the best of the situation because the only choice was to get a different job.

The occasional “bad” passenger is only in my vehicle for a limited amount of time, after the trip is completed I simply give them a star rating of three-stars or less and the software algorithms insure I will never be matched with them again… wish my other jobs had come with this really positive feature!

I guess I understand the “dehumanized” part from the referenced article but with respect:

“Come On Man?!!!  Really?” 

When I take the bus I always say “thank you” to the driver as I exit… if I were operating a commuter light-rail train I’d be up in the front all by myself and wouldn’t get the occasional humanizing positive interaction with passengers.

The nature of the ridesharing “gig” is picking up passengers where they are and taking them to where they want to be… some passengers will go out of their way to acknowledge me as a fellow human being… others won’t say much more than “hello” and “thank you” at the end of the ride.

Some passengers are going to make value judgements because you are a Lyft or Uber driver… but their judgements really say more about them?  They certainly aren’t personal; they aren’t about you the driver.

In short, I believe feeling “dehumanized” because I choose to earn income as a rideshare driver will always be my choice… and I don’t feel dehumanized at all (in fact this idea never occurred to me until I read the referenced article) and I don’t miss the forced nature of being social with co-workers.

I still have lunch almost every week with a co-worker from a job I had years ago… I choose my friends and don’t feel the need to have an employment situation in order to make new friends.  If I did feel this way, I would supplement my rideshare driving income with employment likely to help me meet new friends.

Okay I know I’m going long with this blog and while editing this one my webmaster is going to be shaking his head at my wordiness, so I’ll end with this…

In my opinion what this article is highlighting is the reality that it is our nature as human beings to look for what’s wrong with a given situation (FYI - my other college degree has a psychology focus.)

Recognizing what’s wrong is a human survival technic developed during countless years of human evolution and passed along from our ancestors through our genetic code.  By noticing what appears to be “wrong” humans are better able to predict similar situations in the future and make choices that will protect them from harm.

Every one of us has the ability to focus on what “right” with a given situation and in my opinion, there is a whole lot more “right” about earning income as a rideshare driver than what could be perceived as being “wrong.”

Rideshare driving is not for everyone; and it may not be “right” for you… at least not as your only source of income and only source of human interactions “gig.” 

I strongly suggest you take control of your own experiences and make your own choices.  The Uber matching algorithms are far from the only technology in our current world attempting to “motivate” us away from making our own choices.

Okay gotta go, I’m feeling the need to see if I can get a higher score on my favorite phone application game… hmmm… I wonder why?

What People Hate About Being Managed by Algorithms, According to a Study of Uber Drivers

https://hbr.org/2019/08/what-people-hate-about-being-managed-by-algorithms-according-to-a-study-of-uber-drivers

Uber’s Matching Algorithm Optimizes Corporate Profits, Not Driver Satisfaction?