PARIS — According to his rating on the car-sharing website BlaBlacar, Olivier, an occasional driver who wishes to remain anonymous, is a "really friendly doctor who strikes up super interesting conversations." Though all but one of his passengers have given him a five-star rating, Olivier has trouble accepting the lone rebuke. "Those three stars still gnaw at me," he says.

Christophe Duhamel, founder of the French recipe-sharing website Marmiton, remembers that when the website was launched users often removed their recipes when they were poorly rated. "No one understands their genius," he quips. Others hesitate to post at all, out of fear of bad ratings.

In a world of "serial likers," it's hard to cope with negative remarks. They resurrect our worst school-day prototypes, from the boot-licker, ("We spent the day cleaning the apartment because the prospect of a good rating got us excited"), to the scapegoat ("A guy rated us poorly on Drivy because we rated him poorly, and that brought our average down"), and the rebel ("I refuse to enter this rating system").

"Give me a 10"

Annual company ratings, customer service appraisals, user-to-user websites. It's now possible to spend half the day rating and the other half being rated. When a smartphone owner downloads an app, he is pestered and prodded to decide how many stars to give it. Ask Uber drivers how they cope with the pressure of ratings, and they'll answer, "You know, we rate them too." Discreetly, once they've left the vehicle. And when drivers answer a request, the platform gives them the customer's average score.

[rebelmouse-image 27089919 alt="""" original_size="600x800" expand=1]

Monitoring it in the Philippines — Photo: InsightsUnspoken

In a supply and demand market, scores are meant to push people to give their best and allow users to have access to the best services. In reality, the fear of poor ratings is turning our society into a world of permanent obsequiousness. Everyone is running after good ratings. A hotel receptionist recommending a good museum to a customer will add, "You'll remember my evaluation on Travelocity, won't you?"

Heard recently from a salesperson at a top French mobile phone and Internet provider: "You'll receive a text asking you to rate me from 1 to 10. If you give me less than 10, it's useless to me."

When ratings don't help

Officially, ratings are used to inform users to reassure them when they call on the services of a stranger. In 2015, I rented a car using the private rental website Drivy. When parking next to a sidewalk, I damaged a hubcap. With a traditional rental car company, I would probably have returned the car hoping the breach would go unnoticed. But overcome with panic by the idea of being publicly labeled a bad driver, I changed the hubcap and admitted my mistake. "You shouldn't have," the owner said in the end. And, "You won't forget to leave a few words on the website?"

In this way, ratings are not so much to inform future users as for us to have each other by the short hairs. In a system where everyone rates everyone openly, tacit agreements are made. This is what explains the inflation of good grades in American universities, where students also grade their professors. Mutually grade each other, and everyone will get by fine.

Last year researchers from Boston University calculated that the average Airbnb rating is 4.7 out of 5, with 94% of listings rated between 4.5 and 5. On private-contact platforms, they found that there were lots of very good ratings and a very small number of very poor ratings.

"It's a system that teaches people how to work together," Drivy founder and CEO Paulin Dementhon says. Or, more rarely, to tear each other's guts out, especially because it's possible on his website to change a rating that has already been given ("He gave me a poor rating, so I'll give him a poor one"). For that matter, the website's customer service center receives many calls from users upset by their ratings.

Dementhon doesn't much like his website's public evaluation system — transparent, but not necessarily reliable in terms of information. More opaque evaluation systems such as Uber's or Heetch's are more efficient to inform the user but less so to settle problems.

[rebelmouse-image 27089920 alt="""" original_size="422x226" expand=1]

Certain platforms have tried finding a middle ground. Since the summer of 2014, ratings of the renter and the customer on Airbnb are hidden before being published simultaneously, or 14 days after the end of a stay, without the possibility to change it. BlaBlacar has based its "opinion" system on the same method, "for more sincerity." A pioneer of use ratings, eBay has also evolved its rules over time.

Against the tide

The perversion of rating systems isn't unique to the Internet. It also happens in continuing education, where immediately assessing tutors is mandatory. Some educators even calibrate the end of their work groups to be exciting by making the attendees applaud themselves, for instance, just before rating sheets are distributed. Curiously, this boom in permanent rating comes at a time when schools are going against the tide by developing self-assessments, color grades and other tricks to avoid numbers.

It's the same with businesses. Audit giant Accenture announced last year that it was removing the rating system and the annual ranking of its 330,000 employees. It was too much time and money for limited interest. Two years earlier, Microsoft made the same decision.

But the digital world has a different take on this. At least five start-ups have worked on a possible merger of the user ratings from one app to another, allowing for the "portability" of e-reputations, for better or for worse.

Dementhon concedes, "Everyone rating everyone. That's scary, and it's a sensitive topic."