• UberPeople.NET - Independent community of rideshare drivers. It's FREE to be a person and enjoy all the benefits of membership. JOIN US! CLICK HERE

Uber's Matching algorithm

Get out your tinfoil hats, ladies and gentlemen!

Ever wonder how the algorithm matches drivers with passengers (pax)? One of the frequent lies Uber tells everyone is that the ride request goes to the closest driver. But as experienced drivers know, that’s simply not true. It’s time to take off the hats and look at the actual facts, as (silently) published by Uber.

1568490108930.png


Uber brags about their technical prowess. Recently I stumbled across one of the patents they filed that pertains to their matching algorithm. Credit to @SuperStar3000 for posting the link to this information.

1568490122827.png


I have read the entire thing from end to end. Fascinating reading, really. Once you strip away the lies and half truths that Uber tells the public, as well as the fears, beliefs and imaginations that drivers have, this one piece really tells a more direct story.

First of all, there is not a single thing in there that indicates that drivers wishes are given any consideration when a match is considered. The algorithm works only for the pax, trying to give them the most satisfactory experience that it can. It doesn’t give a rat’s ass about the driver’s satisfaction with various pax, with various rides, with pickup locations or distance, or with anything. It’s all about the passenger and their happiness.

Drivers may be rewarded in general for high AR, but the good news is that drivers are not punished, as many believe, for declining specific rides. The driver’s history and behaviors, however, are recorded, and can (and will) be used for or against them. Lemme ‘splain….

1568490141315.png


Ratings? Yes, your rating makes a difference. Furthermore, your ‘destination rating’ provides a significant bit of influence. For example, if you often get high ratings from people that you take to the airport, you are more likely to be given airport rides. If you get low ratings from people headed to the mall, you are less likely to get rides to the mall. The pickup location is irrelevant; it’s all about the destination.

Ongoing complaints stack up against you. Do you get complaints from pax? If we are not deactivated we often treat them as irrelevant, however they go on our permanent record and are considered when getting matched.

Reputation data and background information is a factor that no one ever thinks of. How’s your credit score? Been arrested recently? Having trouble keeping a long term job in your past? Ever been in the military? Do you associate with suspected gang members or criminals? Divorced? Kids? What church do you belong to? Drop out of college? What kind of political crap do you post on Facebook or Twitter? All of this is factored in somehow into your matching, believe it or not!!!

Punctuality is another one I hadn’t expected. Once you get a ping, how long until you arrive? I take the punctuality score to mean how often do you arrive as predicted. Do you screw around for a minute before heading out? If you have a 12 minute pickup and are going slow to get that “Pickup Premium”, your 16 minute arrival may be costing you future pings.

Don’t believe me? Think the tinfoil hat is on too tight? Look at the picture above one more time and read it again for yourself!

Oh, but there’s more! Read on…

1568490156994.png


Pax rating (of drivers) and their satisfaction level on previous rides is recorded and analyzed when matching with future drivers. So if a pax rates well when matched with an outspoken multiracial female member of the LGBTQ community, who's under the age of 30, and both a Muslim and a Democrat, drives a super-clean newer mid-sized vehicle and lives on the East side of town, Uber is going to try and match them with as many of the same possible categories in the future. The more rides the passenger takes, the greater the likelihood of the algorithm finding the right combination of driver/ride/vehicle factors that the pax wants.

If a pax values a driver’s rating, those of you who have high ratings are more likely to get those rides. If a pax values energy efficient vehicles, those of you with Priuses are more likely to get those rides. If a pax gave five stars to the last 98 out of 100 drivers with green eyes, then that will be considered as a criteria. Smokers seem to get more drivers who happen to smoke, isn’t that amazing?

It’s not just one variable, it’s many, and it’s based on the unknown and random preferences of each individual pax.

There was a time when I got 5 rides in 2 weeks from people who were visiting from Atlanta, Georgia. I thought it was strange, such an odd coincidence. Now it makes more sense.

Your actions, vehicle cleanliness, manner of driving, etc are all being monitored. Do you always drive the speed limit? If so, you are more likely to be matched with people who prefer that. Do you listen to talk radio while pax are in the car? You will likely get pax who have the same political beliefs or who just don’t care, maybe because they wear headphone during rides. Do old people love you? Do hood rats think you’re cool? It’s not about what YOU prefer, it’s about who prefers YOU, and why.

1568490182794.png


I can appreciate the discrimination when it comes to vehicle type, that makes sense to me. But country of origin? Home address? Eye color? Propensity towards alcohol? Age? Affiliation with scandalous groups? Delinquent financial behavior?

We criticize Uber for failure to do background checks. But really, they are digging further into our backgrounds than we know! Everything that gets posted on FB, Instagram, LinkedIn, Twitter, etc etc etc is part of our ‘reputation data’. If you post pictures of your dog on FB or volunteer at the local animal shelter, I suspect you are more likely to get service dogs than the rest of us.

Matching isn’t so random after all. Do any of these things shout DISCRIMINATION to anyone else besides me? Uber stopped short of saying that race is a factor, however they indicated that the above items are only a partial list of items. It would be illegal for them to admit that they factor race into the equation, but I strongly suspect that race is a major criteria.

And yes, this has been happening to all drivers for years, and continues. Day after day after day.

Whatever happened to “pax are matched with the closest driver”?

1568490194530.png


Check out the top part – they give the example of a surge taking place, and if a particular pax is likely to give a low rating (because they are paying too much); the AI will try to match them with someone who will overcome that money issue and “earn” a five star rating from the pax anyway.

The bottom part sums up this entire article: The matching algorithm tries to match pax with drivers based on the hope that the pax (not the driver) is 100% satisfied with the ride.

Drivers: You will get more potential matches with a high AR from both companies, but the belief of the app punishing drivers for rejecting a ride from Walmart is false. ‘Karmic matching’ (taking a crap ride in order to get better rides in the future) is nothing more than a myth. The bottom line is that the AI uses blatant discrimination in order to give the individual pax the best experience they possibly can, based on their history and their personal information, as well as yours.

So much of what we write in UP is opinion. I thought I would make this one about factual published information. Hope it helps to open some eyes. Feel free to share your comments and thoughts.
 
Mr T

Comments

ariel5466

Well-Known Member
Great article, and it makes sense. I'm a 29 year old female and I get A LOT of female pax in my age range. I'm also white and English is my first and only language, and I've gotten more than a few pax who don't like immigrants. I've chalked it up to coincidence but you cited the proof that it isn't. Makes me wonder if the algorithm could put me in danger by matching me with someone who prefers young female drivers for nefarious purposes.
 

Cassiopeia

Active Member
I find it frightening to think that corporations are making programs that decide who goes where and with whom. Your blog gives good insight into our corporate computer controlled reality.

I’m sure AirBnB approaches their algorithm similarly in many ways. There are a lot of systems making decesions for us every day that we do not notice.
 

Invisible

Well-Known Member
This is extremely informative and thought provoking. It explains why I’ve been getting so many pax who recently moved here. Last week, I had 3 pax in a day from CA. I get so many pax from Chicago as well. But the algorithm doesn’t care that for me some of the Chicago pax are the worst!

Makes me wonder if the algorithm could put me in danger by matching me with someone who prefers young female drivers for nefarious purposes.
I hope not for your sake! However, you seem like you can whip any guy’s ass.
 

dirtylee

Well-Known Member
Yeah, I've noticed this over time as well. I tend to mostly get business pax during the day or solo females at night. Like back to back to back.

The destination thing is true as well, I think it takes into account what the driver rate at the end. I always rate areas of town I hate to go 1 star no matter what & go offline. I get those areas a lot less over time. Around here, the local uber office resets pax ratings anyways.

The algorithm shenanigans go out the window when it surges highly.
 

Wolfgang Faust

Well-Known Member
Get out your tinfoil hats, ladies and gentlemen!

Ever wonder how the algorithm matches drivers with passengers (pax)? One of the frequent lies Uber tells everyone is that the ride request goes to the closest driver. But as experienced drivers know, that’s simply not true. It’s time to take off the hats and look at the actual facts, as (silently) published by Uber.

View attachment 356233

Uber brags about their technical prowess. Recently I stumbled across one of the patents they filed that pertains to their matching algorithm. Credit to @SuperStar3000 for posting the link to this information.

View attachment 356234

I have read the entire thing from end to end. Fascinating reading, really. Once you strip away the lies and half truths that Uber tells the public, as well as the fears, beliefs and imaginations that drivers have, this one piece really tells a more direct story.

First of all, there is not a single thing in there that indicates that drivers wishes are given any consideration when a match is considered. The algorithm works only for the pax, trying to give them the most satisfactory experience that it can. It doesn’t give a rat’s ass about the driver’s satisfaction with various pax, with various rides, with pickup locations or distance, or with anything. It’s all about the passenger and their happiness.

Drivers may be rewarded in general for high AR, but the good news is that drivers are not punished, as many believe, for declining specific rides. The driver’s history and behaviors, however, are recorded, and can (and will) be used for or against them. Lemme ‘splain….

View attachment 356235

Ratings? Yes, your rating makes a difference. Furthermore, your ‘destination rating’ provides a significant bit of influence. For example, if you often get high ratings from people that you take to the airport, you are more likely to be given airport rides. If you get low ratings from people headed to the mall, you are less likely to get rides to the mall. The pickup location is irrelevant; it’s all about the destination.

Ongoing complaints stack up against you. Do you get complaints from pax? If we are not deactivated we often treat them as irrelevant, however they go on our permanent record and are considered when getting matched.

Reputation data and background information is a factor that no one ever thinks of. How’s your credit score? Been arrested recently? Having trouble keeping a long term job in your past? Ever been in the military? Do you associate with suspected gang members or criminals? Divorced? Kids? What church do you belong to? Drop out of college? What kind of political crap do you post on Facebook or Twitter? All of this is factored in somehow into your matching, believe it or not!!!

Punctuality is another one I hadn’t expected. Once you get a ping, how long until you arrive? I take the punctuality score to mean how often do you arrive as predicted. Do you screw around for a minute before heading out? If you have a 12 minute pickup and are going slow to get that “Pickup Premium”, your 16 minute arrival may be costing you future pings.

Don’t believe me? Think the tinfoil hat is on too tight? Look at the picture above one more time and read it again for yourself!

Oh, but there’s more! Read on…

View attachment 356236

Pax rating (of drivers) and their satisfaction level on previous rides is recorded and analyzed when matching with future drivers. So if a pax rates well when matched with an outspoken black-haired single female Chinese Muslim Democrat lesbian under age 30 that drives a super-clean newer mid-sized vehicle and lives on the East side of town, Uber is going to try and match them with as many of the same possible categories in the future. The more rides the passenger takes, the greater the likelihood of the algorithm finding the right combination of driver/ride/vehicle factors that the pax wants.

If a pax values a driver’s rating, those of you who have high ratings are more likely to get those rides. If a pax values energy efficient vehicles, those of you with Priuses are more likely to get those rides. If a pax gave five stars to the last 98 out of 100 drivers with green eyes, then that will be considered as a criteria. Smokers seem to get more drivers who happen to smoke, isn’t that amazing?

It’s not just one variable, it’s many, and it’s based on the unknown and random preferences of each individual pax.

There was a time when I got 5 rides in 2 weeks from people who were visiting from Atlanta, Georgia. I thought it was strange, such an odd coincidence. Now it makes more sense.

Your actions, vehicle cleanliness, manner of driving, etc are all being monitored. Do you always drive the speed limit? If so, you are more likely to be matched with people who prefer that. Do you listen to talk radio while pax are in the car? You will likely get pax who have the same political beliefs or who just don’t care, maybe because they wear headphone during rides. Do old people love you? Do hood rats think you’re cool? It’s not about what YOU prefer, it’s about who prefers YOU, and why.

View attachment 356237

I can appreciate the discrimination when it comes to vehicle type, that makes sense to me. But country of origin? Home address? Eye color? Propensity towards alcohol? Age? Affiliation with scandalous groups? Delinquent financial behavior?

We criticize Uber for failure to do background checks. But really, they are digging further into our backgrounds than we know! Everything that gets posted on FB, Instagram, LinkedIn, Twitter, etc etc etc is part of our ‘reputation data’. If you post pictures of your dog on FB or volunteer at the local animal shelter, I suspect you are more likely to get service dogs than the rest of us.

Matching isn’t so random after all. Do any of these things shout DISCRIMINATION to anyone else besides me? Uber stopped short of saying that race is a factor, however they indicated that the above items are only a partial list of items. It would be illegal for them to admit that they factor race into the equation, but I strongly suspect that race is a major criteria.

And yes, this has been happening to all drivers for years, and continues. Day after day after day.

Whatever happened to “pax are matched with the closest driver”?

View attachment 356238

Check out the top part – they give the example of a surge taking place, and if a particular pax is likely to give a low rating (because they are paying too much); the AI will try to match them with someone who will overcome that money issue and “earn” a five star rating from the pax anyway.

The bottom part sums up this entire article: The matching algorithm tries to match pax with drivers based on the hope that the pax (not the driver) is 100% satisfied with the ride.

Drivers: You will get more potential matches with a high AR from both companies, but the belief of the app punishing drivers for rejecting a ride from Walmart is false. ‘Karmic matching’ (taking a crap ride in order to get better rides in the future) is nothing more than a myth. The bottom line is that the AI uses blatant discrimination in order to give the individual pax the best experience they possibly can, based on their history and their personal information, as well as yours.

So much of what we write in UP is opinion. I thought I would make this one about factual published information. Hope it helps to open some eyes. Feel free to share your comments and thoughts.
Best post in months.
Kudos, T.
tmobilep.jpg
 

lyft_rat

Well-Known Member
Great summary, Mista T. Proximity will still carry a lot a weight under certain conditions, but as a "cherry picker" it explains why I keep getting my regular customers even when I am not so close due to another trip.
Post automatically merged:

I love getting a ping when I know who the person is and where they are going.
 

iheartuber

Well-Known Member
I still believe timeouts are real. How is it that when I cancel or decline a ride I “coincidentally” don’t get another ping for 2-30 min? The only saving grace is if I’m in a busy or surge zone

I used to explain it like this: if Uber needs drivers, the timeout goes out the window. But if there’s ants everywhere, you get the “punishment”
 

GreatWhiteHope

Well-Known Member
Get out your tinfoil hats, ladies and gentlemen!

Ever wonder how the algorithm matches drivers with passengers (pax)? One of the frequent lies Uber tells everyone is that the ride request goes to the closest driver. But as experienced drivers know, that’s simply not true. It’s time to take off the hats and look at the actual facts, as (silently) published by Uber.

View attachment 356233

Uber brags about their technical prowess. Recently I stumbled across one of the patents they filed that pertains to their matching algorithm. Credit to @SuperStar3000 for posting the link to this information.

View attachment 356234

I have read the entire thing from end to end. Fascinating reading, really. Once you strip away the lies and half truths that Uber tells the public, as well as the fears, beliefs and imaginations that drivers have, this one piece really tells a more direct story.

First of all, there is not a single thing in there that indicates that drivers wishes are given any consideration when a match is considered. The algorithm works only for the pax, trying to give them the most satisfactory experience that it can. It doesn’t give a rat’s ass about the driver’s satisfaction with various pax, with various rides, with pickup locations or distance, or with anything. It’s all about the passenger and their happiness.

Drivers may be rewarded in general for high AR, but the good news is that drivers are not punished, as many believe, for declining specific rides. The driver’s history and behaviors, however, are recorded, and can (and will) be used for or against them. Lemme ‘splain….

View attachment 356235

Ratings? Yes, your rating makes a difference. Furthermore, your ‘destination rating’ provides a significant bit of influence. For example, if you often get high ratings from people that you take to the airport, you are more likely to be given airport rides. If you get low ratings from people headed to the mall, you are less likely to get rides to the mall. The pickup location is irrelevant; it’s all about the destination.

Ongoing complaints stack up against you. Do you get complaints from pax? If we are not deactivated we often treat them as irrelevant, however they go on our permanent record and are considered when getting matched.

Reputation data and background information is a factor that no one ever thinks of. How’s your credit score? Been arrested recently? Having trouble keeping a long term job in your past? Ever been in the military? Do you associate with suspected gang members or criminals? Divorced? Kids? What church do you belong to? Drop out of college? What kind of political crap do you post on Facebook or Twitter? All of this is factored in somehow into your matching, believe it or not!!!

Punctuality is another one I hadn’t expected. Once you get a ping, how long until you arrive? I take the punctuality score to mean how often do you arrive as predicted. Do you screw around for a minute before heading out? If you have a 12 minute pickup and are going slow to get that “Pickup Premium”, your 16 minute arrival may be costing you future pings.

Don’t believe me? Think the tinfoil hat is on too tight? Look at the picture above one more time and read it again for yourself!

Oh, but there’s more! Read on…

View attachment 356236

Pax rating (of drivers) and their satisfaction level on previous rides is recorded and analyzed when matching with future drivers. So if a pax rates well when matched with an outspoken black-haired single female Chinese Muslim Democrat lesbian under age 30 that drives a super-clean newer mid-sized vehicle and lives on the East side of town, Uber is going to try and match them with as many of the same possible categories in the future. The more rides the passenger takes, the greater the likelihood of the algorithm finding the right combination of driver/ride/vehicle factors that the pax wants.

If a pax values a driver’s rating, those of you who have high ratings are more likely to get those rides. If a pax values energy efficient vehicles, those of you with Priuses are more likely to get those rides. If a pax gave five stars to the last 98 out of 100 drivers with green eyes, then that will be considered as a criteria. Smokers seem to get more drivers who happen to smoke, isn’t that amazing?

It’s not just one variable, it’s many, and it’s based on the unknown and random preferences of each individual pax.

There was a time when I got 5 rides in 2 weeks from people who were visiting from Atlanta, Georgia. I thought it was strange, such an odd coincidence. Now it makes more sense.

Your actions, vehicle cleanliness, manner of driving, etc are all being monitored. Do you always drive the speed limit? If so, you are more likely to be matched with people who prefer that. Do you listen to talk radio while pax are in the car? You will likely get pax who have the same political beliefs or who just don’t care, maybe because they wear headphone during rides. Do old people love you? Do hood rats think you’re cool? It’s not about what YOU prefer, it’s about who prefers YOU, and why.

View attachment 356237

I can appreciate the discrimination when it comes to vehicle type, that makes sense to me. But country of origin? Home address? Eye color? Propensity towards alcohol? Age? Affiliation with scandalous groups? Delinquent financial behavior?

We criticize Uber for failure to do background checks. But really, they are digging further into our backgrounds than we know! Everything that gets posted on FB, Instagram, LinkedIn, Twitter, etc etc etc is part of our ‘reputation data’. If you post pictures of your dog on FB or volunteer at the local animal shelter, I suspect you are more likely to get service dogs than the rest of us.

Matching isn’t so random after all. Do any of these things shout DISCRIMINATION to anyone else besides me? Uber stopped short of saying that race is a factor, however they indicated that the above items are only a partial list of items. It would be illegal for them to admit that they factor race into the equation, but I strongly suspect that race is a major criteria.

And yes, this has been happening to all drivers for years, and continues. Day after day after day.

Whatever happened to “pax are matched with the closest driver”?

View attachment 356238

Check out the top part – they give the example of a surge taking place, and if a particular pax is likely to give a low rating (because they are paying too much); the AI will try to match them with someone who will overcome that money issue and “earn” a five star rating from the pax anyway.

The bottom part sums up this entire article: The matching algorithm tries to match pax with drivers based on the hope that the pax (not the driver) is 100% satisfied with the ride.

Drivers: You will get more potential matches with a high AR from both companies, but the belief of the app punishing drivers for rejecting a ride from Walmart is false. ‘Karmic matching’ (taking a crap ride in order to get better rides in the future) is nothing more than a myth. The bottom line is that the AI uses blatant discrimination in order to give the individual pax the best experience they possibly can, based on their history and their personal information, as well as yours.

So much of what we write in UP is opinion. I thought I would make this one about factual published information. Hope it helps to open some eyes. Feel free to share your comments and thoughts.
Damn this is super interesting
 

GammaRayBurst

Well-Known Member
It's weird I get a lot of single females at night in the middle of the night I guess female passengers don't feel unsafe with me specifically white females and asian females it's very strange when my rating went up two points I started to notice that I started to get more airport runs further away from me because I know I'll drive 20 miles to pick up somebody I don't care I get $20 per trip minimum and lately I've been noticing that I get tips on every single airport run which I never used to get tipped all the time I noticed that I changed my style slightly but I've been doing this for months I drop somebody off at the airport and then I drive 20 miles to the rich areas only and I've noticed that my clients are all rich which I find strange because even during the daytime I get rich clients now I've been getting tip more than ever I guess they prefer me even though I'm not a professional I noticed that during bar run won't believe this but I get mainly drunk females then than drunk males I noticed this about 2 months ago and im a male! I do go over the speed limit slightly so I don't know if that's good or bad for me but when I'm on a trip I'll go the speed limit I've noticed now that going the speed limit though I've been getting higher payouts and richer people I guess they want to feel very safe and won't rate Five Stars unless you go the speed limit
 

lyft_rat

Well-Known Member
It's weird I get a lot of single females at night in the middle of the night I guess female passengers don't feel unsafe with me specifically white females and asian females it's very strange when my rating went up two points I started to notice that I started to get more airport runs further away from me because I know I'll drive 20 miles to pick up somebody I don't care I get $20 per trip minimum and lately I've been noticing that I get tips on every single airport run which I never used to get tipped all the time I noticed that I changed my style slightly but I've been doing this for months I drop somebody off at the airport and then I drive 20 miles to the rich areas only and I've noticed that my clients are all rich which I find strange because even during the daytime I get rich clients now I've been getting tip more than ever I guess they prefer me even though I'm not a professional I noticed that during bar run won't believe this but I get mainly drunk females then than drunk males I noticed this about 2 months ago and im a male! I do go over the speed limit slightly so I don't know if that's good or bad for me but when I'm on a trip I'll go the speed limit I've noticed now that going the speed limit though I've been getting higher payouts and richer people I guess they want to feel very safe and won't rate Five Stars unless you go the speed limit
WHOA you gotta go to punctuation class. Cool name though, GammaRayBurst.
 
Top