The Sound of Silence, or What Uber and Airbnb’s Missing Data Tells Us

Episode 1: May 2014

Last May, Uber wrote this in a blog post:


UberX driver partners are small business entrepreneurs demonstrating across the country that being a driver is sustainable and profitable.  For example, the median income on uberX is more than $90,000/year/driver in New York and more than $74,000/year/driver in San Francisco.

There are all kinds of issues with the claim, but put them aside for the moment. Instead, here’s a question: assuming the claim to be true, what does Uber’s post tell us about the income of uberX drivers in cities other than New York and San Francisco?

One obvious thought is that it tells us nothing, because it only makes claims about NYC and SF. Another is that we could extrapolate from these numbers as if they were randomly chosen, and guess that the income in other cities is somewhere in a distribution centred around $82K. That’s what the Washington Post did by implicitly comparing the Uber numbers to a national average for taxi drivers:

Estimates of the typical cab driver’s salary hover around $30,000. According to Uber, the median wage for an UberX driver working at least 40 hours a week in New York City is $90,766 a year. In San Francisco, the median wage for an UberX driver working at least 40 hours a week is $74,191.

But Uber’s blog post actually tells us something different about driver income in other cities: it tells us that it is less than $74,000. How come? Simply because this is a blog post from Uber. Uber wants to make a compelling case to the media and to the public, so if drivers in Chicago or Los Angeles were making more money than those in NYC or SF they would have appeared in the post instead of those two cities. So we don’t know how much drivers elsewhere make, but we can be pretty sure that it’s less than $74,000.

Episode 2: Oct 2014

In October 2014, Uber published another blog post saying that drivers in New York City take an average of $36.16 per hour in fares, and make $25.17 per hour after Uber’s cut, but before expenses. (That’s about $50,000 per year for a 40-hour week, by the way.) What does this tell us about driver income in other cities? Again, it tells us that it’s probably less than $25 per hour because otherwise they would not choose New York City for their blog post.

Episode 3: January 2015

So last week Uber posted both a survey of its drivers and a report (PDF) by eminent economist  Alan Krueger and Uber Head of Policy Research Jonathan Hall. The main things to come out of the report were that Uber is growing really quickly – impressive, but no surprise — and that Uber drivers are paid better than taxi drivers. That conclusion comes from Table 6:  uberwages

The table confirms what I was saying above: earnings in New York City are 50% more than anywhere else apart from San Francisco, and San Francisco is the second highest earning city.

As several other commentators have pointed out (eg Ellen Huet at Forbes, Jacob Davidson at Time, Andrea Peterson at the Washington Post), the comparison between Uber “earnings per hour” and taxi “hourly wages” is apples and oranges, because the report has nothing to say about Uber driver expenses. Krueger and Hall have this to say in their defence:

A detailed quantification of driver-partner costs and net after-tax earnings is a topic of future research. Nonetheless, the figures suggest that unless their after-tax costs average more than $6 per hour, the net hourly earnings of Uber’s driver-partners exceed the hourly wage of employed taxi drivers and chauffeurs, on average.

(Oh Alan Krueger: you were given a set of data and “full discretion over the content of the report” but you come out with this? I’m not angry, I’m just disappointed.)

So what do we know about driver expenses? The Sound of Silence rule tells us that, if expenses made a good story for Uber, expenses would have included expense data in the data set given to Krueger. In fact, if expenses made a good story, they would have been included in the earlier company posts as well. So we can be pretty sure that the expenses of drivers are $6/hour or more in most cities.

Is $6/hour expenses reasonable? Certainly: the Washington Post’s Peterson notes that the IRS estimate of costs is 57.5 cents per mile, but that Uber don’t include distance driven in their report. Still, 12 miles an hour (including driving to pick up a fare and driving back to a good spot afterwards) is not implausible. And then there is the thorny question of commercial insurance, which would be an additional cost. It seems likely to me that Uber drivers are getting about the same as taxi drivers.

Episode 4: January 2015

Just to finish off, here is another sound of silence. Thanks to the efforts of Share Better, my data on Airbnb rentals got picked up in New York City last week when City Council held hearings about short term rentals. That data suggests that the percentage of illegal rentals in Airbnb’s New York listings has decreased only slightly since a year ago, and that the decrease has been swamped by an increase in the overall number of listings. There is not enough firm data on the Airbnb public web site to be absolutely solid on this, but that’s how things look.

What did Airbnb have to say? Their response is that the data is “flawed” and “inaccurate”, and that is literally the extent of what they have to say. So it’s kind of a joke. This piece in Mashable by Jessica Plautz includes Airbnb saying they don’t really know anything about their hosts, and Josh Dzieza in The Verge includes more of Airbnb’s exchange with the council in which the company says, well, nothing.  Airbnb executive David Hantman’s testimony ‘acknowledged that his company did not keep track of the number of users who are renting out their apartments illegally. “We don’t research that,” he said during the hearing.’

Sometimes I wonder if I’m doing the right thing going after Airbnb, because there is a role for occasional casual short-term rentals. Maybe the company’s heart-on-its-sleeve attitude is for real? But then I realize that in the last year they could have put out realistic, convincing numbers at any time (and saved me a lot of work) but they haven’t. And like Uber’s silence on driver expenses, Airbnb’s silence speaks volumes: if the data were good for the company, they would have made them public. The fact that they haven’t (and their content-free responses to people trying to piece together a picture of reality) makes me more convinced than ever that they have things to hide.



The 2001 Anthrax Deception

I may be biased, but I believe my friend Graeme MacQueen has written a brave and important book in his newly-published “The 2001 Anthrax Deception“, which he subtitles “The case for a domestic conspiracy”. It’s a book that takes on new importance in the light of the CIA torture report.

If you’re like me, you won’t have thought about the anthrax attacks in some time. Remember: these were the letters containing spores of anthrax sent through the US mail to media companies and to selected politicians in the immediate aftermath of the 9/11 attacks; letters that killed five people. During October 2001 there was a widespread belief that the letters were a follow-up to the 9/11 planes, carried out by al-Qaeda with the help of Iraq, which was supposedly one of the few actors with the expertise and capacity to produce this biological weapon of mass destruction. That theory was disproved in late October and the FBI began hunting for a domestic “lone wolf”, initially pursuing one Steven Hatfill and then, when this avenue also proved to be a dead end, going silent. Six years later the FBI and the Department of Justice announced that they had a new suspect in Dr. Bruce Ivins, who worked on anthrax vaccines for the US army. Ivins died – said to be suicide – shortly before he was to be charged. In 2010 the Department of Justice closed the case, releasing a report that laid the blame at Ivins’s door.

MacQueen’s book (I think of it as “Graeme’s book” of course, but for the sake of style I’ll call him “MacQueen” here). As I was saying, MacQueen’s book is brave because any attempt to question official stories of events surrounding 9/11 get labelled as “conspiracy theories” which are, after all, taboo in polite society. Questioning of the official story of the anthrax attacks has been made to seem tantamount to a belief in the Illuminati and a conviction that the moon landings were faked, but MacQueen refuses to back away from the word. It’s to his credit that he puts the word “conspiracy” right there on the cover. And after all, as he says, it is nearly universally believed that the 9/11 attacks were a conspiracy.

It should be uncontroversial to question the “official story” of the anthrax attacks. There was no trial, no opportunity for a defence by the man deemed guilty; the official story has not been tempered by the heat of public interrogation and debate. Released years after the events, the appetite for challenging the report has never been a strong one. In short, as official stories go, it’s not well established; its status is more like an arrest than a conviction.

And people do conspire, MacQueen reminds us. To use the word need not imply a belief in all-powerful yet shadowy cabals, nor commit us to a conspiratorial view of history; the spectrum of possibilities is broad. Maybe it helps to think of a more distant case. The death of the Princes in the Tower during the Wars of the Roses remains something of a mystery, but most historians lay the blame at the feet of Richard III. Yet to say that “Richard did it” is just the beginning. There’s almost no chance that Richard actually carried out the execution with his own hands. So did he order it directly? Or did he utter an equivalent of the earlier Henry II’s “will no one rid me of this turbulent priest?” and let his councillors take their cue from his frustration? Or did the initiative come from elsewhere in his court, as someone lower down the hierarchy took history into their own hands? All of these are possible. There is a supply-and-demand aspect to political entrepreneurship and to responsibility for heinous acts. Creating the conditions that encourage and reward criminal actions is part of the tangle of blame.

We know that sending anthrax powder through the mail is not so far from confirmed practices of US military and intelligence agencies: Oliver North arranged for arms to be sold to Iran in order to pursue war in Nicaragua through contra proxies; the Gulf of Tonkin incident demonstrated the willingness of the US to deceive the public in order to justify a war. The use of agent provocateurs by official agencies is well established: many of the terrorist attacks supposedly prevented by anti-terrorist agencies were facilitated and encouraged by operatives of those same agencies as they try to establish their radical credentials in the communities they are infiltrating. Here in Canada, Mubin Shaikh was actively involved in setting up training camps that were then used to justify the arrest of the youths who attended them.

All of which is simply to say that there is a case to be made here, and one that should be well within the norms of acceptable debate.

One of the reasons I came into the book in a sceptical frame of mind was that there are no planning documents, no insiders giving compelling accounts of how the planning was done and where. Someone, I thought, would have come out and said something; or someone would have found and leaked some key document. But the report on CIA torture puts those doubts to one side: it shows that illegal actions can be carried on with the involvement of many inside people, that even though there would be tremendous demand for such material from the press no one leaked anything over the years since 2002, and that in the absence of the inquiry we would still not know (as a society) that these events had happened and that the CIA had organized them. So yes, secrets can be held for years.

In all this waffle I realize I’ve not actually said much about the book itself, so maybe I should summarize it at least. Here goes.

MacQueen sets out his aims right at the beginning of the book. He’s out to show that the perpetrator of the anthrax attacks was not a “lone wolf”, that they were carried out by deep insiders in the US executive branch, that there were links to the 9/11 Hijackers, that the goals of the attackers were to redefine the enemy of the west to the Global War on Terror and to weaken the rule of law internationally and within the US: goals that were successfully achieved. He convinces me of the first and of the final two. I’m not quite ready to follow him for sure with the second and third, but he makes a compelling case for both and I am convinced by him that insiders within the US executive branch have responsibility for these attacks, whether or not there was an explicit order to carry them out.

The focus of the book, as the subtitle says, is to make the case for a domestic conspiracy, and MacQueen starts with motive. In September and October 2001 there was a concerted effort to cast the events of 9/11 as an act of war rather than just terrorism. Casting it as war was essential to passing the Patriot Act, to giving the President extraordinary powers, and to setting up military tribunals to try those seized. The case for war was made much stronger by the anthrax attacks, which clearly required capabilities the reach of non-state actors, and which were presented (during the crucial weeks of debate during October 2001) as clear evidence of conspiracy between al-Qaeda and Iraq. If Iraq was involved, then it’s an attack by a nation state, hence war.

In the lead-up to passing the Patriot Act on October 11, 2001, the Bush administration adopted a “strategy of tension”, building up the threat of an al-Qaeda/Iraq axis. Leahy and Daschle played leading roles in delaying the Patriot Act so that it did not pass by an administration deadline of October 5th and then, between October 6 and 9, letters were sent to both of them containing anthrax. Sure enough, on October 11 the act was passed. When it was signed into law on October 26, George Bush specifically invoked the anthrax attacks as a justification for curtailment of civil rights, in a chain of events that led straight to the now-exposed NSA surveillance practices. There is no doubt that the anthrax attacks gave the Bush administration exactly what it needed in order to achieve its legislative goals.

The central chapter of the book is an investigation of the four possibilities for the attacks: (domestic/foreign) (individual/group). The early focus was, as I’ve already said, on a “foreign group” and it’s important to remember just what happened at that time. On the basis of flimsy and ultimately false information, conjecture, and circumstantial evidence (at best), there was a major part of the US (and international) political and journalistic establishments who proclaimed, with sufficient certainty to invade countries and kill large numbers of people, that the verdict was in and the attacks were the responsibility of a “double perpetrator”: al-Qaeda and Iraq. It is at the very least inconsistent to dismiss those, like MacQueen, who suggest that a domestic group perpetrator was responsible when those who propagated and used the “double perpetrator” hypothesis continue to enjoy prestige and respect. The “conspiracy theorists” have a right to be taken seriously in this debate.

The double-perpetrator hypothesis gave way in the face of the revelation that the anthrax was weaponized and featured an aerosol additive that indicated it came from US military laboratories. A foreign individual was clearly out of the question, so that leaves two options. All efforts were devoted to tracking a domestic individual, first Steven Hatfill and then Bruce Ivins, efforts that concluded with the publishing of the Department of Justice report in 2010 and the closing of the case.

The main focus of the book is not to attack the case against Ivins, but MacQueen does highlight two events: the publishing of the National Academy of Science report in 2011 that said the physical evidence tying Ivins to the mailed spores was inconclusive, and researchers who published scientific papers showing that the spores could not have been prepared by an individual, and probably came from a lab other than the one where Ivins worked.

I found this central chapter very strong. The case against Ivins is far from compelling, the absence of a trial is remarkable, and there has simply been no public attempt to assess guilt or innocence. MacQueen makes a very strong case that there is no obviously true official story, and that attempts to establish alternative theories are valid and should be taken seriously.

I was less convinced by the following chapter, which is about advance knowledge of the attacks. Certainly there was much discussion of the possibility of anthrax attacks, and of dirty nuclear bombs, in the media during September 2001, after 9/11 and before the actual anthrax attacks became public in October, but I’m not convinced that such discussion necessarily derived — as MacQueen suggests it does –from the perpetrators of the attacks or that it showed foreknowledge or an attempt to prepare the ground for when the attacks actually happened.

That said, there is, again, a spectrum of possibilities. MacQueen’s claim that top officials used the chatter and amplified it at politically useful times and that the chatter was part of a conspiracy to justify war is convincing, as is his claim that the double perpetrator hypothesis was “false, known to be false, and flowed from repeated acts of deception by the US Government”.

If anything does make me consider the possibility of foreknowledge it’s MacQueen’s account of an exercise called Dark Winter, carried out at Andrews Airforce base on June 22-23 2001. The exercise simulated a biological warfare attack by terrorists in which anonymous letters containing smallpox virus are sent to media outlets. The simulation evolved in striking parallel to the actual events of September and October 2001. The participants in Dark Winter included Judith Miller (New York Times journalist who played an important role in promoting war on Iraq), James Woolsey (director of the CIA under the Clinton administration) and Jerome Hauer (a member of the neoconservative Committee on the Present Danger and also the director of the Office of Emergency Management of New York City until early 2000).

What do these parallels tell us? They don’t prove that the team organizing the Dark Winter exercise carried out the anthrax attacks (and MacQueen doesn’t claim this), but they do tell us that there was a well-established script which gives the attacks meaning and purpose from a conservative point of view. They show that the people involved in Dark Winter, together with others in the Bush administration and in the conservative media, fostered a climate in which the anthrax attacks played a politically useful role, creating a context which encouraged a “false attack” by anyone who would want to see an attack on Iraq happen. Such an attack took place, and played the part required of it. Whether it was actually executed by a lone wolf or by a group is perhaps not the most important question.

MacQueen is interested in the anthrax attacks for their own sake, but he is also interested in 9/11, and he sees connections between the two. Robert Stevens, the first casualty of the anthrax attacks, was an employee of Mike Irish at Boca Raton newspaper The Sun and a friend of his wife, real-estate agent Gloria Irish. Gloria Irish found houses both for Stevens and, in the summer of 2001, for two of the 9/11 hijackers (Marwan al-Shehhi and Hama al-Ghamdi). The official explanation for this is that it’s a coincidence. Beyond this, Hijacker Mohamed Atta made attempts to buy crop-dusting planes which made sense only for an anthrax attack (which al-Qaeda had, we now know, no capability to execute). It does seem likely that there was an attempt to tie the anthrax attacks to the 9/11 attacks, and so to present al-Qaeda as likely anthrax attackers – but by whom? Bruce Ivins could not have known (MacQueen shows) of the connections between Stevens and the Hijackers.

The Bush administration undertook a wide range of deceptive actions – lying, making false implications and more – to present anthrax, al-Qaeda and Iraq as a powerful and threatening combination. These actions culminated in Colin Powell’s UN Security Council presentation at which he held up a vial of simulated anthrax. So just how far did they go? How much did they push, cajole, and encourage their organizations to produce evidence to back up the stands they were taking? And if they were putting a lot of pressure on their organizations to provide the proof of al-Qaeda/Iraq mendacity that they knew was out there, would it be surprising if a group somewhere undertook to provide the nudge that was needed, or took the pragmatic shortcut of just doing it themselves? Because surely, you can imagine them thinking, even if this attack was fake, it would be done with the best of motives, to draw attention to a greater danger that they knew was there in Baghdad. Will no one rid me of this turbulent priest?

The book gets, for me, at least 80% of the way to the author’s goals. He reminds us of the lengths to which powerful organizations go to achieve their aims. He reminds us that such organizations are not all-powerful, but make mistakes. He shows us that the constellations were aligned by the Bush administration in such a way that the anthrax attacks (by domestic actors supportive of the Bush administration goals) would be an understandable action. He shows us that there is a plausible connection between the perpetrators of the anthrax attacks and the 9/11 Hijackers. If he doesn’t pull me the last mile to being convinced that the attacks were executed by people deep within the US executive branch I am a lot more open to the possibility than when I started. It’s a brave and necessary book and it matters that dissident voices such as his are demanding that questions continue to be asked about this important recent history.

Why Canada should de-activate Uber

A long post collecting together a number of arguments and resources about Uber and why it doesn’t belong in Canada. If you want to print it off, here is a PDF.

An updated version of this page is maintained here, which you can access through the menu above.

In September, Uber introduced its UberX service in Toronto as part of a broad expansion across Canada (among other countries), and the City of Toronto is now seeking a court injunction to stop the company. Back in December 2012 the company was chased out of the city, but now some journalists, business professors, and even the mayor (not that one) think it’s time to disrupt the taxi industry, ignore the city licensing and standards process and let Uber run its taxi alternatives.

You’d think, reading these articles, that customer service is the only issue at stake here. But there’s more to it than that: the decision about Uber is also a decision about the kind of jobs we want to have, about accessibility, and about the kind of city in which we want to live. It’s about us as consumers, but also us as citizens, us as Canadians, and us as employees.

Uber is not “the future”, it’s “a future”

Contrary to the way some articles are written, we do have a choice here. A lot of the links above talk as if Uber were some kind of inevitable future (“The Ubers are destined to win the taxi wars”); here is an example from Todd Hirsch:

It’s an economic tale told time and again. Camera film makers. Video stores. The music recording industry. Perhaps most famously, the Luddites – those textile labourers in 19th century England who protested against the introduction of mechanized looms by smashing them. Many of them failed to adapt to new, disruptive technologies and went extinct.

Next on the list may be the taxi industry…

Conflating Uber with the broad advance of technology is just wrong, and it’s also exactly what Uber wants us to do. After all, if you’re the future, who can argue against that? But thousands of new technology businesses start every year, and many of them fail. Groupon turned out not to be the future of shopping. Mayor John Tory wants to “sit down with the Ubers and the Hailos and others of the world”, but Hailo has already folded its North American operations. Paris has turned down UberPop, but it has Autolib’, which may be the world’s most successful electric-car sharing program. There are many roads to the future—many innovative roads to the future—and the best of them don’t involve Uber.

The Uber controversy is not just—or even mainly—a technology story, it’s fundamentally a deregulation story; the story of a uniquely American fundamentalist free-market worldview being sold to us in the name of “car-sharing” and innovation.

Two cheers for regulation

Uber CEO Travis Kalanick’s says that his company is engaged in something like an election race in which

Uber is the candidate and [its opponent] is an asshole called Taxi. I’m not totally comfortable with it but we have to bring out the truth of how evil Taxi is.

Kalanick also referred to “Our opponent – the Big Taxi cartel” when he hired former Obama strategist David Plouffe to run its political lobbying efforts.

It’s worth reflecting for a moment on the fact that cities all around the world, with many different political and economic traditions, have decided independently that taxis need some regulation; and there is no “Big Taxi cartel” coordinating these decisions. Taxi firms are generally city-wide, at least until Uber itself came along.

Cities have had their own reasons for regulating taxis. As McGill University Law Professor Paul Stephen Dempsey wrote in a 1996 paper, regulations typically include:

(1) limited entry (restricting the number of firms, and/or the ratio of taxis to population), usually under a standard of “public convenience and necessity,” [PC&N] (2) just, reasonable, and non-discriminatory fares, (3) service standards (e.g., vehicular and driver safety standards, as well as a common carrier obligation of non-discriminatory service, 24-hour radio dispatch capability, and a minimum level of response time), and (4) financial responsibility standards (e.g., insurance).

It is not a coincidence that, all around the world, taxi regulation happens at the city level rather than at a national or provincial level. Different cities have different needs. The one-size-fits-all model that Uber wants to roll out treats Tokyo the same as London the same as Windsor, but as Dempsey (see link above) writes:

In the final analysis, the suitability of taxicab service and pricing is a peculiarly local issue, best tailored by local governments based on their unique populations, spatial densities, road congestion, air pollution, and airport and hotel traffic.

Meanwhile, for all its attempts to conjure up an underdog image, Uber has funding from (among others) Google, Jeff Bezos of Amazon, and Goldman Sachs.

It’s also worth remembering that taxi deregulation has been tried and failed in many places. For example, Seattle deregulated in 1979 but found that “service quality declined and rates were often higher”; and a 2004 US report noted that

…several studies, including a 1993 Price Waterhouse study, found that overall, in many cities that deregulated, the supply of taxicabs increased, fares increased, service quality declined and there were more trip refusals, lower vehicle quality, and aggressive solicitation of customers resulting from a higher supply of taxicabs.

So taxi regulation is imperfect, but the solution to that imperfection is not to walk away from it. Taxis are part of city infrastructure, like buildings, buses, and subways, and serve a valuable public service. Taxi regulation provides a way for the public and for our elected representatives to have a say in how the taxi industry works, so that (for example) Toronto can demand that the entire taxi fleet must be wheelchair accessible, rolling in over the next decade, or London can introduce a zero-emission Metrocab to address environmental concerns. The taxi industry may be a flawed part of our democratic institutions, but the solution to flawed democratic institutions is more and better democracy, not the American rejection of government and of democracy’s role in the economy.

Despite claims of technological inevitability, saying No to Uber is perfectly compatible with innovation and forward thinking. If Canadian cities such as Toronto do say No to Uber, they would be joining cities around the world such as Berlin, Seoul, Madrid, and Paris in refusing to roll over and hand the responsibility for their taxi service to Silicon Valley.

Uber corporate culture

Digital platforms are governed by network effects, which is why Uber’s massive funding rounds (having raised $1.5 Billion, they are now looking to raise another $1 Billion) and rapid global expansion are so important to the company’s future. They and their investors know it’s a winner-take-all game; that’s why Uber CEO Travis Kalanick admitted this month that he undermined the fund-raising efforts of his main competitor, Lyft. The decision facing Canadian cities is not about permitting a smartphone app, it’s about handing a large part of Toronto’s taxi service over to Uber.

So what kind of company is Uber, that people want to put aside taxi regulation and hand control over to them?

It’s a company in the middle of controversy, as executive Emil Michael muses about digging up dirt on (female) journalists who criticize the company. It’s a company that spies on its customers using what it calls its “God View” of company data for party tricks or for blog posts. It’s a company whose top New York exec is being investigated internally for tracking a female journalist without her consent. It’s a company whose employees have warned another female journalist that “company higher-ups might access my rider logs”.

I emphasize “female journalist” here because the culture of the company is a big part of the problem. Do we want to hand over Toronto streets to a company whose CEO wisecracks about women on demand (“Yeah, we call that Boob-er“)? Which “can and does track one-night stands” and posts a blog entry about the data called “Rides of Glory” (now removed)? Which, more damagingly, ran a campaign in France (now also removed) called “Avions de Chasse” to pair Uber riders with “hot chick” drivers? As the company posted on the English version of its website:

“Avions de chasse” is the French term for “fighter jets”, but also the colloquial term to designate an incredibly hot chick. Lucky you! the world’s most beautiful “Avions” are waiting for you on this app. Seat back, relax and let them take you on cloud 9!

This is a company that responded to a complaint by a female customer that she was driven 20 miles out of her way to an abandoned lot by saying it was an inefficient route.

It’s also a company that coaches Miami drivers how to circumvent laws. It’s a company that ordered and cancelled over 5,000 rides with its main competitor to interfere with their service. It’s a company that sets out to dupe newspapers with fake PR, and which now talks of “weaponizing facts” in its PR campaigns (this company really likes its military analogies). In short, it’s a company that reflects the worst of the macho culture of some parts of the technology industry.

In among all the privacy problems, it is worth noting that Uber has one privacy policy for the US, but has no Canadian privacy policy. That’s how much Uber is thinking, for all its massive funding base, about privacy issues for Canadian riders.

Drive income: the Uber unicorn

What really appeals to people who have used Uber (and I am not one of them) is the smooth efficiency of the service, an efficiency that relies on drivers to be in place, ready to take you where you want to go.

According to Uber, efficiency and affordable pricing is combined with fantastic pay for drivers. In May 2014 the company posted a claim on its website that the median uberX annual income is $90,766 in New York City and $74,191 in San Francisco. The claim was hailed (hah!) by Matt McFarland of the Washington Post with the headline Uber’s remarkable growth could end the era of poorly paid cab drivers. Noting that “estimates of the typical cab driver’s salary hover around $30,000″, McFarland acknowledged that “Uber’s numbers don’t account for the costs a driver incurs to own and operate a vehicle. Still, the gap in compensation for providing similar services is astounding”.

The claim is indeed astounding. How could Uber pay its drivers three times the pay of a taxi driver, charge riders less, and still be profitable? The initial explanation was that taxi medallion owners in many cities suck all the money out of the taxi system (see financial journalist Felix Salmon here and here for example).

I took a look at the claim when it was made and compared Uber’s estimates to taxi reports from three different cities. It was clear that Uber was taking out about the same percentage of each dollar of fare from the system as do medallion owners (20% of the fare at the time, plus a $1 “safety fee” per ride), so there was no magic there after all. The company admitted that it was looking only at drivers who logged over 40 hours a week which, given that most drivers are working longer than they are actually “on the app” means that the average is highly selective of the top end of drivers. Finally, Uber massively underestimated the costs of gas, maintenance, insurance, and other expenses (such as tolls) that drivers incur (uberX drivers must drive their own cars).

Since then, the claim has become less and less plausible. Felix Salmon toned down his praise here. Seattle Uber drivers protests in April and August, San Francisco drivers protested low income in May, Los Angeles drivers protested in September, New York drivers protested in September (Uber backtracked over new rules it had introduced) and again in October along with San Francisco and London. Low income has been a consistent complaint, including on the active Uber driver forum UberPeople and at Reddit. Reports from individual Uber drivers failed to come near to the income figures that Uber claimed (for example this report from Fort Worth (strong language!) or this report or this report from Boston).

Meanwhile, Uber has been taking a bigger and bigger slice of the consumer dollar. The $1 per trip “safety fee” was a start, then they started charging drivers $10 per week for use of a smartphone, and more recently they increased their commission for new drivers in San Francisco to 25% of the fare.

Finally, multiple rounds of fare cuts have put pressure on driver income. Uber insists that a greater number of rides per driver makes up for the price cuts, but drivers themselves dispute this claim and there are few numbers from Uber to back it up.

The $90,000 claims have been re-investigated in recent weeks by two of the most consistently informative journalists covering Uber. Johana Bhuiyan of Buzzfeed looked at eleven Uber drivers’s pay slips here; and Slate’s Alison Griswold called the $90,000 driver Uber’s Unicorn, saying “In several months of reporting on Uber, I have yet to come across a single driver earning the equivalent of $90,766 a year.” She asked Uber’s Lane Kasselman to introduce her to one and concludes: “Last I heard, they’re still looking.” Meanwhile, the New York Post concludes that “their take-home pay is closer to the average that yellow cabbies make.”

Rating systems and vulnerable drivers

Money is, of course, one of the main points of contention for many jobs, but Uber is not just another employer. In fact, it’s not an employer at all: Uber drivers are “partners”, self-employed entrepreneurs who choose to work on the platform. The model of “micro-entrepreneurs” who can choose when to work independently is what makes Uber part of the booming “sharing economy” along with others such as Airbnb. What seems at first like a light-weight and flexible model of work turns out, in Uber’s hands, to be another way for the company to have its cake and eat it too.

Uber claims its drivers are not employees, but has been exerting more and more control over their behaviour.

Uber makes it easy for new drivers to join up: it publicizes what many drivers describe as unrealistic earnings to attract interest (see above) and encourages drivers to take out what amount to subprime loans in order to buy a car. Then, of course, the driver has to put in long hours to pay off the purchase, and being kicked off the platform becomes even more of a threat.

Uber has taken advantage of the vulnerability of its drivers by imposing more and more strenuous rules. Driver acceptance rates (when the app assigns them to a customer) have to be up around 90% or you get a notification to “Please improve your acceptance rate if you want to continue to use the Uber platform”. Drivers claim to have been deactivated for being critical of the company on Twitter. Drivers on the premium Uber Black service (I have mainly ignored the complexity of the different Uber services here) were suddenly forced to take requests for the lower pay uberX service. The company tracks driver locations and complains if they are not to the company liking.

But at the heart of the control is the rating system that permits riders to rate drivers. Most riders give their drivers five stars out of five as a courtesy, but if a driver’s rating slips even slightly – below 4.7 in many cities – they can be “deactivated” or kicked off the platform. The system makes drivers vulnerable to the most demanding of riders, as a small number of complaints can lead to the driver losing their way of making a living. And of course there is no appeal, as the driver isn’t an employee and the contract is not a work contract. The reports of happy and friendly Uber drivers (see the Globe and Mail articles at the beginning of this post) take on a different meaning once you know the precariousness of Uber drivers’ situation. As Forbes’ Jeff Bercovici reported, “Uber likes this system because it enjoys being able to say all of its drivers have near-perfect ratings. But it’s a harsh one for drivers, and also for customers, who find themselves repeatedly forced to choose between guilt, spite and ignorance.”

While Uber gets to dictate the behaviour of its drivers in more and more specific ways, it still takes none of the responsibility when things go wrong. Section 230 of the “Communications Decency Act” may seem like an odd law to protect the company, but here’s how it works. The law was initially introduced to say that blog sites and other user-content sites such as YouTube were not responsible for content posted by its users. Fair enough. But now Uber says it’s not a taxi company, it just runs a web site and an app, and puts drivers in touch with riders. Anything that goes wrong is not Uber’s responsibility, it’s the driver’s. The law is an American one, but challenging Uber is going to be expensive and maybe prohibitively so for Canadians too, especially given the company’s formidable bank account.

Uber’s rules seem to step over the line regarding whether a driver is, or is not, an employee according to Canada Revenue Agency rules. Sharing economy workers are facing this issue with other companies too, such as cleaning/odd-job service Handy. It’s an issue that other industries face, such as construction, and the root cause is always the same: classification as an independent contractor relieves the hiring company (Uber in this case) from having to pay EI premiums, and from having to abide by employment standards. The risk is pushed entirely onto the subcontractor. When Toronto mayor John Tory says that Uber should be allowed to operate, he’s implicitly approving of these arrangements, and allowing bad labour practices to intrude further and further into Canada’s workplaces. It’s bad for the drivers, and it’s bad for our society.

Customer issues

Among all of these problems, the customer is the one who does well (putting aside the problems with sexism and privacy violation listed at the top of the post). Uber is keeping the prices low, pushing drivers to accept all the rides they are assigned whether or not it makes them money, and is using the rating system to ensure that the drivers put on a friendly face. It’s relying on these customers to push cities into permitting the apps.

But even here there are concerns: issues of race are complex: some people of colour report a dramatically better experience than with taxis but others are concerned about customer profiling, and potential for discrimination. Disability access is more troubling: blind riders are suing Uber and disability activists are filing lawsuits.

Summing up

This post is already too long. I could have written about the occasional horror stories of accidents and abusive behaviour from drivers, but they happen with taxis too. And I’ve not covered the dubious screening process that Uber uses, or the lack of proper mechanical inspection of the cars, or the difficulties with insurance: these have all been covered elsewhere.

The list of problems with Uber is long. The existing taxi industry is not perfect, especially in big cities, but it is a part of the democratic fabric of the country, and we need to address its problems by better democracy, not by throwing democratic accountability aside for the promise of self-regulation by an unaccountable company with an arrogant, sexist corporate culture, which treats its drivers badly, which walks away from its responsibility when things go wrong, and which is importing an American anti-democratic attitude into this country.

Lucky Me

Three writerly things I’m enthusiastic about at the moment.

First, my recent post In Praise of Fake Reviews has been reworked and polished thanks to the efforts of Rob Horning, and posted at The New Inquiry.

Second, I’m off to New York City next week to take part in the Digital Labor conference. I’ll be on a panel on The Future of Workers in the Sharing Economy; we each talk for about 30 minutes and then discuss. I’ll be talking in part about reputation.

Third is an article in progress, but more on that later.



In Praise of Fake Reviews

(4000 words on why arranging for fake reviews can be a legitimate response to reputation systems such as Yelp, where businesses cannot opt out, and where the site owner is neither accountable nor widely trusted.)

The story of Botto Bistro

You may have seen the campaign by the owners of Botto Bistro to become the worst-reviewed restaurant on Yelp; it’s brilliant, it’s successful, and it has significance way beyond Richmond just north-east of San Francisco, where Botto Bistro has been serving customers for years.

Radio Free Richmond was the first outlet to mention the campaign, and that was on September 12. Three days later the story was picked up by Mike Aidax in the Richmond Standard, and two days after that by Paolo Lucchesi of the San Francisco Chronicle. Then it went national, with appearances on CNN and CNBC. The most comprehensive account I’ve seen was by Cyrus Faravar in Ars Technica on September 21, which I’ll draw on again and again in what follows.

But let’s go back to the beginning. It was five years ago that Davide Cerretini and Michele Massimo quit their jobs to set up a restaurant in a “worse location—a ghetto—and serve amazing food without expectation of customer service or tablecloths.” And they seem to have been true to their word, adopting a take-it-or-leave-it attitude to their customers: “I work 12 hours a day, six days a week. But if you come here to try to make us change whatever it is that is in our place, just shut up and go somewhere else.”

So “Botto Bistro serves all of its food at its order counter—there’s no table service—and crams 10 tables amidst giant flat-screen TVs” and their website FAQ is awesome, with gems like this:

Q. Do you serve pizza by the slice?
A. No.
Q. Why?
A. Still no.

It sounds like something from Fawlty Towers, and you can imagine it’s not for everyone, but Davide and Michele have built a clientele and are apparently doing fine.

Enter Yelp

So why the trouble with Yelp? Well, according to Cerretini, Yelp has been continually pestering the restaurant to advertise on the web site “30 times a week, sometimes five times a day”. This may be an exaggeration. Still, for six months, Botto Bistro did advertise. Then they cancelled their ads but Yelp kept calling. “I insulted them and I called them bad names”, says Cerretini, “Yelp asked us to pay them as advertising so they can criticize me better.”

In frustration, the Botto Bistro owners flung up their hands (metaphorically, I wouldn’t want to stereotype Italians) and started their Hate us on Yelp campaign to become the worst-reviewed restaurant in the San Francisco area. Thirteen hundred reviews later, they proudly announced a one-star aggregate rating on Yelp, the lowest rating possible.1

Their Yelp page is a lot of fun. Some reviewers compete for the most outrageous complaint (“Their pizza ruined my marriage”, “I’m sitting here in my office in Hamburg/Germany and are in need of some food and learn that you are not able to deliver???”, “WORST place EVER!” and so on). Others are more obviously supportive (“Had this same idea when Yelp screwed over my business… just never went about doing it! Very glad to see it working out for you.” “Fuck you, Yelp. Long live Botto.”) Everyone is having a good time.

But Yelp has, stupidly enough, responded, as Faravar tells us. On September 15 “Marvin” from Yelp sent one of the co-owners an email that started this way:

Hi Davide,

I ‘m contacting you from the Yelp User Support Team because we’ve received complaints from the community that you may be offering incentives in exchange for reviews.

To be clear, this violates our Terms of Service…

After accusing Davide of violating Terms of Service to which he has not agreed, the email goes on to threaten “a Consumer Alert being placed on your listing”.

So three cheers for Botto Bistro. Its campaign is many things: it’s a gimmick, a gesture of frustration, a declaration of individuality and quirkiness, and an act of rebellion against the creeping tyranny of a world defined by software and Terms of Service.

But the campaign relies on novelty for its success, so it’s not a tactic that can be easily reproduced by others. What, then, can other restaurants do if they are frustrated by Yelp? It’s clear that some resort to “fake reviews”, with less panache than Botto Bistro. It’s an act that’s widely seen as both ethically and legally wrong, but maybe there is a place for it…

What’s wrong with fake reviews?

We all know what’s wrong with fake reviews. Crowd-sourced sites such as Amazon’s book reviews, TripAdvisor’s hotel ratings, and Yelp’s restaurant recommendations, depend for their success on being the honest and authentic opinion of those who have read, stayed at, or eaten at the subject of their review. If restaurant owners reward people, or pay people, to submit positive reviews (sometimes people who have not even eaten at the restaurant) then those reviews are “frauds”: Yelp talks about “fake” or “phony” reviews, and calls them “attempts to deceive consumers”.

Some law enforcement agencies agree that paying for reviews qualifies legally as fraud: last year New York regulators carried out a sting operation that got “19 companies to cease their misleading practices and pay a total of $350,000 in penalties.”

“What we’ve found is even worse than old-fashioned false advertising,” said Eric T. Schneiderman, the New York attorney general. “When you look at a billboard, you can tell it’s a paid advertisement – but on Yelp or Citysearch, you assume you’re reading authentic consumer opinions, making this practice even more deceiving.”

The finding of “false advertising” was not based on whether the content of the reviews matched reality, but on the fact that the reviewers were posing as authentic customers, while actually being paid. But of course there is nothing unusual in a business paying somebody to speak positively about their offering. We’ve all seen television advertisements featuring celebrities who are paid large amounts of money to endorse a product or service.

The claim that it’s worse when it’s not on an obvious billboard is also outdated. We all know that TV and movie product placements are endorsements inserted precisely to seem natural – where you can’t “tell it’s a paid advertisement” – by actors who may never have used the product they are endorsing. We’ve heard about “viral marketing” stunts in which people are paid to act as genuine enthusiasts for a product. This may be a bit sleazy, but it is also just business as usual.

There are many places where advertisement and “authentic” comment is getting mixed up. It’s of real concern where the venue is a public venue: the practice of drug companies promoting their own brand of painkillers in university lectures is an obvious one, but I’m sure you can think of more.

But why, at a time when ads and “authentic” comment mix in Facebook, Twitter and YouTube, is the Yelp reputation system off-limits as an advertising medium? Why is it “false” to pay people to say nice things about your business on Yelp’s review listings but not “false” to solicit endorsements on LinkedIn (tip: my LinkedIn profile is almost entirely fake), in the middle of a prime time television show? Why should Yelp reviews not be just another venue for marketing and promotional activities?

Yelp and restaurants: incentives

Just as Google’s search results pages hold both paid-for advertising listings and “organic” listings, so Yelp’s pages contain both advertising and “real reviews by real people”. Yelp contends that the two are kept separate, but the potential for conflict of interest is obvious and some business owners believe that Yelp manipulates its listings to benefit businesses that advertise with them. And I’ll come back to that later.

But Botto Bistro’s frustration with Yelp is different. The owners are not interested in promoting themselves on the site. They want nothing to do with Yelp. The catch is, they can’t opt out.

Here is one of Yelp’s Common Questions:

Can I have by business listing information removed from Yelp?

Consumers have the right to talk about what they like (and don’t like) about a meal they ate, a plumber they hired, or a car wash they visited. We don’t remove business listings, so your best bet is to engage with your fans and critics alike, and hear what they have to say.

(Aside: what a patronizing tone Yelp adopts as it tells business owners how to run their own enterprises, and how unsurprising that Yelp recommends actions that benefit—who else?—Yelp.)

If you’re a restaurant owner, Yelp wants three things from you. First, it wants reviews of your business in order to draw people to its site when they are looking for a place to eat. Second, they want to make money from you by selling you ad space. Third, they want to profit from the reviews of your business and your ads by striking partnerships with Microsoft, Yahoo, and maybe other search engine providers. It’s going to get two of those whether you like it or not.

The restaurant’s compulsory appearance on Yelp is both a promise and a threat. Yelp is the steward of the restaurant’s presentation on the site and paying Yelp is a way to improve that presentation. Regardless of whether Yelp actually manipulates its listings, there is an obvious conflict of interest in the roles.

If it’s going to be compulsory for restaurants to appear on Yelp, then we have to ask questions about the company’s accountability before we judge those who pay for reviews or (even worse) charge them with fraud. And the picture is not pretty…

While restaurants are held to strict standards when it comes to how they behave on Yelp, it turns out that Yelp itself is not accountable to the law at all. (I’ll look at American law here: if you’re a restaurant outside the US then holding Yelp accountable will be even more challenging).

It’s well known that Section 230 of the Communications Decency Act protects web site owners from being responsible for user-provided content on their sites, even when that content is the source of profits. The protection extends to Yelp, according to a judge in 2011 and reaffirmed in September 2014. Yelp has no legal responsibility to verify any comments made by reviewers, for factual accuracy, reasonableness, or anything else. So if a restaurant complains that a reviewer’s opinion is not based on fact, that’s not Yelp’s problem. Opinions can be both “authentic” and wrong.

But of course Yelp does not just assemble user content, it also arranges it and presents it. Yelp’s sorting algorithms can make a big difference to a restaurant’s business: is the company bound to be fair in its presentation of restaurants? Business owners recently took Yelp to court claiming that the company “extorted advertising payments from them by manipulating reviews on its site”. A US Federal Appeals Court found that—whether or not Yelp is manipulating reviews—it can do what it wants.

…even if small business owners could prove that Yelp manipulates their pages to persuade them to pay up for advertising, that doesn’t mean they are being extorted, or that Yelp is violating federal law… the law does not require Yelp to publish positive reviews in the first place, Judge Berzon wrote.

“By withholding the benefit of these positive reviews,” the judge wrote, “Yelp is withholding a benefit that Yelp makes possible and maintains. It has no obligation to do so, however.”

So Yelp is fundamentally unaccountable for the fairness of its reputation system.

The usual argument is that it is in Yelp’s own interest to maintain a trustworthy site—that market forces will keep it honest. But while it’s clear that a completely spam-filled site would lose its value, it is not clear that the optimal amount of “review fraud” from Yelp’s point of view is zero. Yelp is interested in volume as well as accuracy: it encourages participation by reviewers, restaurants, and users through a whole gamut of techniques. There is always going to be a balance between this encouragement and dissuasion of so-called “biased” reviewers. We cannot trust loosely-defined market forces to preserve fairness.

Yelp as a protection racket

So Yelp is not accountable, but what about its actual history? Does it act as a steward or is it running a protection racket?

It is known that Yelp asks for and gets free services in exchange for reviews. Each year Yelp rewards some of its reviewers by anointing them part of the “Yelp Elite Squad”, and among the benefits are some “epic parties“. So who pays? Here is what Yelp’s official blog says:

Sometimes Yelp foots the bill directly for the venue, food or drink; oftentimes we look for sponsors to help defray those costs (e.g., Peroni Beer comped the beer I drank last week in New York). The venues themselves typically provide the space, some staff and sometimes food or drink. This arrangement is great for yelpers because we can visit a new place and meet each other affordably. And it works well for the venues and sponsors because they get exposure to a group of local adventurers who, if they like what they see, may become future customers.

There’s a clear quid-pro-quo here: free stuff for reviews. Not that it’s explicit, but of course it doesn’t have to be. Ironically, Yelp understands how the implicit quid-pro-quo works when it comes to the small businesses being reviewed: here is an excerpt from Yelp’s advice pages for businesses:

Why does Yelp discourage businesses from asking for reviews?

Would-be customers might not trust you. Let’s face it, most business owners are only going to ask for reviews from their happy customers, not the unhappy ones. Over time, these self-selected reviews create bias in the business listing – a bias that savvy consumers can smell from a mile away. No business is perfect, and it’s impossible to please 100% of your customers 100% of the time.

Business owners obviously feel the implicit promise when approached by Yelp. According to the Chicago Tribune, back in 2009 Chicago cupcake shop owner Patty Rothman was approached by a Yelp staffer who “walked into her Gold Coast shop and ‘guaranteed us good reviews on the site if we catered one of their parties for free’. Offended but resigned, Rothman complied. And just as promised, positive reviews bloomed for the business right after the party, Rothman said.”

Or Doug Quint and Bryan Petroff of Big Gay Ice Cream in New York, which opened in 2011:

As Quint and Petroff explain, last summer before their New York City shop even opened its doors for the first time, they got an email from a so-called “Elite” Yelper that read, “I am going to bring a team of other Elite Yelpers on an ice cream crawl and we want to come to your store. Can you open early?” And then, in addition to asking the Big Gay Ice Cream guys to come up with special flavors just for them and sample-size portion of the whole menu, this Elite Yelper went in for the kill: “Think of all the great reviews you’ll get.” (link)

Beyond the solicitation of free services, accusations of the Yelp advertisement sales team employing protection racket tactics over the ratings posted on its site are widespread. Here are just a few:

From the LA Times in August 2013, reporting on a town hall for business owners:

…when the town hall opened up for Q&A an hour and a half later, business owners were quick to vent their frustrations.

Many slammed the company for allowing review posters to write inflammatory comments – one restaurant manager said she cried for three days after a Yelper said her restaurant was filled with Nazis, another said he’d received negative reviews from Yelpers who admitted they’d never tried the food – while others said they had been subjected to aggressive advertising calls from Yelp.

“I have one-star reviews for my diner from people that have never walked into the place. They’ve never stepped foot, they’ve never tried the food, but they give me one-star reviews. That’s insane. Why would you let someone like that stay on the site?” demanded Craig Martin, owner of Cafe 50s. “I spoke to your office, I called your guys, I emailed, I talked to your salespeople.”

Another business owner, vintage clothing shop owner Reiko Roberts, said the advertising pressure amounted to extortion. She said that when she declined to buy ads, “the lower reviews go to the top and the higher reviews go to the bottom.”

From a Reddit Ask Me Anything session with the CEO, here are two of many:

My anecdotal experience: I told the Yelp sales rep to get lost and to stop calling me after about 6 months of hounding. He finally relented… but guess what ALL of my positive reviews are now filtered. Up until that conversation, only about 25% of my reviews were filtered. Now EVERY SINGLE ONE.

I sat in on a sales call with a Yelp rep, and after questioning her on the finer details we decided Yelp wasn’t for us and told her so. She then became noticeably irritated and said “Good luck getting any positive reviews in the future.”

And for more, you can look (where else), on Yelp’s page on Yelp, where it has a rating of 2.5 and 3105 one-star reviews.

You might wonder if these claims are true. Is it fair of me to post these allegations without checking into their accuracy?

Well, one reason restaurant owners get angry at Yelp is that Yelp does not care whether factual claims made by reviews are true or verifiable: their rankings of businesses are based on opinion, regardless of veracity.

So I’m happy to tell you that my sophisticated, but unfortunately proprietary, algorithms have identified all these reports of Yelp as honest opinions of the people involved. I’m sure you can find Yelp’s response to these allegations elsewhere, but they haven’t created an account here on Whimsley, so I’m not going to post a link. What’s sauce for the goose is sauce for the gander.

I don’t see that a business owner has any ethical responsibility to maintain the integrity of a system that they feel is threatening their livelihood, is run like a protection racket, and which is unresponsive to their appeals. Seems to me like treating Yelp as a venue for marketing is a perfectly reasonable course of action.

Why trust reputation: bigger questions

So that’s the point of the essay, but before I finish let’s explore another issue. Part of the problem is Yelp itself, but if Yelp were better, would there still be a case for fake reviews? Sometimes yes.

There are problems that come from the design of the reputation system itself, even if it’s run “fairly”.

Yelp aggregates user ratings to show an overall “score” for each restaurant and then it ranks the restaurants in order of this score; Yelp uses this “aggregate and rank” approach to decide which restaurants to highlight and which to regulate to the bottom of a list.

Yelp is proud of the influence of its model. When Cyrus Faravar asked Yelp’s vice president of corporate communications Vince Sollitto about Botto Bistro, Sollitto replied: “I think that you have to wonder if this is the smartest strategy—there’s a Harvard Business School study that was done showing that for every [star] increase, restaurants see an uplift in revenue between five and nine percent.”

Yelp would maintain that a good restaurant will get good ratings from its customers and so will benefit from the site, but the story is not so simple.

While the net effects of reputation systems are complex and varied, there is a homogenizing, winner-take-all component that comes from this practice of ranking preferences and displaying them in order of average value. Yelp heavily favours those businesses that appear in the top few of the listings.

Yelp’s decision to rank based on average reviews favours restaurants with a broad appeal, and acts against more polarizing venues. It’s the same reason that many challenging or controversial movies that divide audiences (“cult movies” belong in this class) score lower on Rotten Tomatoes than those that are less divisive. Cult movies may attract a more intense following among their core audience, but there is nowhere in the Rotten Tomatoes system for such intensity to be reflected.

Like a cult movie, Botto Bistro’s owners set out to deliver their own vision in an uncompromising fashion. Appeasing those who don’t like it is not high on their list of priorities. As they told Faravar: “We specifically, aggressively inform people that if you’re looking for customer service, this is the last place you should be.”

There is a place for many types of restaurant, including those with Botto’s take-it-or-leave-it style. But it’s clear why Botto Bistro will never get far on Yelp even if they wanted to: Yelp’s whole model of reputation works against the idiosyncratic and difficult.

The homogenizing force of “average and rank” systems may become more severe as whole industries seek to increase the role of digital ratings in our daily lives. There are many in the technology world who dream of reputation as currency: the idea that a numerical reputation score on one system (such as Yelp) should be portable to other environments (TripAdvisor? Credit Card companies? Landlords?) A good score on eBay may indicate a “good reputation” that you could parlay into a better interest rate on a bank loan. It’s already happening to some extent, as Airbnb posts the number of Facebook Friends of each host as evidence of their trustworthiness. This kind of naive scoring system is a terrible idea. The only thing worse would be a sophisticated scoring system.

It’s well known that early reviews can have a big effect on many ratings systems by “setting the tone” for reviews that follow. If you know that early reviewers have given a restaurant five stars but you are underwhelmed, you may give it three stars rather than two.

The aggregation process itself can also affect your business. A couple of years ago Rebecca Rosen reported in The Atlantic on a study by Michael Anderson and Jeremy Magruder: the researchers seized on the observation that two restaurants can have very similar average ratings (say 3.24 and 3.26) and yet when Yelp rounds them into half-star categories one will be rounded down to 3.0, while the other is rounded up to 3.5. Looking at 400 restaurants in the San Francisco area, they discovered that “moving from 3 to 3.5 stars is associated with being 21 percentage points more likely to have sold out all 7:00 PM tables and moving from 3.5 to 4 stars makes restaurants an additional 19 percentage points more likely to have sold out all tables.” That is, Yelp’s own choices about how to present ratings, independent of business “quality” reflected in its average rating had a significant effect on some businesses.

In short, the impact of almost any reputation system on a restaurant’s business is not just the natural effect of that business’s quality, but is also a result of the system designer’s choices, which cannot be influenced by the business owner. Business owners are dependent, to some extent, on the design choices of an unaccountable company for their success or failure. That’s not a pleasant position to be in and it’s made worse by the “winner-take-all” structure of many social media marketplaces. If a few fake reviews are needed to limit the impact of large reputation systems, that’s fine with me.

The philosophy of ratings

Let’s finish up with what philosophers call “the question of meaning.” If we are to believe in a “fair reputation system” we have to ask: what, grasshopper, does a one-star (or a five-star) review really mean?

Yelp’s view (and TripAdvisor’s, and Netflix’s, and Amazon’s, and Airbnb’s, and eBay’s … you get the picture) is that a rating is an indication of individual preference, and that aggregating these preferences can be used to recommend the experience to others. If “reputation” is the social distillation of others’ opinions about ourselves, then these rankings are a measure of “reputation”, distilled multiple times into a single number.

But just because that’s what the aggregators would like a rating to mean doesn’t make it so. There is a world of context behind each click that may not fit into any algorithm’s assumptions. In Botto Bistro’s case, a one-star review is a gesture of support, a way to thumb your nose at an impersonal system, and a quick path to cheap pizza. It’s a dramatic example of how one “one-star review” may have a completely different meaning from another “one-star review” even though they are recorded in Yelp’s systems in the same way.

It’s tempting to think that if only “fake” reviews could be removed, then the remaining ones would indicate preference, but there’s little to back up such a claim. In other contexts (like eBay or Airbnb) a five-star review is more often a courtesy than a rating. Less dramatically, in the case of Netflix, one outcome of the Netflix prize was that reviews of one movie are heavily primed by what other movies we have recently reviewed (so that “objectivity” is tainted).

And maybe a “fake review” should be seen as a commitment on the part of the restaurant owner, an investment in branding that indicates an intent to stay in the marketplace, a sign that the restaurant can, in fact, be trusted — just like any other investment in branding.

Ratings do not always mean what data scientists would like them to mean. Human expressions are protean, and no attempt by Yelp or others to pin them down will succeed completely. There is something postmodern about ratings, just as any other human expression. We imbue them with meaning and intent outside of what the system wants, and there is no objective fact behind it all that is the “true” rating. Il n’y a pas de hors-click.

Reputation is a multi-faceted, qualitative concept. It has been pushed through a meat-grinder by digital reputation systems and has come out the other side homogenized, devoid of texture, but easier to digest. There’s nothing inherently wrong with reputation systems by themselves, but giving them too much authority or influence will inevitably throw up bad incentives on the part of the system owner and those taking part in the reviews. Before we declare “fake reviews” to be a crime or an obvious case of bad morals, we at least need to demand some accountability from reputation site owners.


1 As of this writing, Botto’s rating on Yelp has gone back up to 1.5, because the company has relegated over 600 reviews to the “not recommended” bin and removed over 800 for violating content guidelines, leaving only 560 reviews on the front page.

Uber Drivers Organizing

A short note about the ways that Uber drivers are getting increasingly vocal and organizing to push back against the company and its frequent rate cuts, policy changes, and no-tip policy. This post lists only American events.


  • In Los Angeles, drivers protested fare cuts. Uber says they will make more money (because of more rides) but never backs that claim up. It’s hard to think why drivers would protest if they were better off.
  • In New York, up to a thousand drivers protested over rule changes, fare cuts, and Uber’s no-tip policy. And it looks like they got Uber to back down on some of their policy changes.
  • The Washington Post has a good piece about the challenges drivers face.
  • After a protest over fare cuts in June, Seattle drivers took to honking their horns in August.

Legal challenges

  • In Boston, Attorney Shannon Liss-Riordan is taking on Uber over its claim that drivers are not employees.
  • In San Francisco, a lawsuit has been given the go-ahead over Uber’s gratuity policy.
  • Meanwhile, a San Francisco ruling that FedEx drivers really are employees (not independent contractors) may have an impact on Uber down the road, as a big part of its business model is to avoid any of the commitments that go with providing actual jobs.

Online Talk

If you want to get a glimpse into how Uber drivers are thinking, there are active forums at and on a reddit subforum. See, for example, this description of driving in LA since the most recent rate cut.