Prime Suspect
Amazon's multi-front battle with fake reviews, labor, data privacy and facial recognition
Happy Tuesday, a.k.a. day two of Amazon Prime Day!
Many of you have likely been busy hunting for deals (mythical beard oil, anyone?), so it felt right to kick things off with “The Everything Store.” This is definitely jumping into the deep end given Amazon’s massive and far-reaching businesses, but the company has been getting a lot of attention recently and it’s worth taking a closer look.
Also, Amazon’s associate general counsel, Nate Sutton, is set to testify before the House Judiciary Committee today as part of its antitrust investigation.
But first, here are a few other stories that are worth keeping tabs on:
Facebook’s cryptocurrency project, Libra, draws scrutiny
Article: Sylvan Lane / The Hill
What happened: Federal Reserve Chairman Jerome Powell told members of Congress last week that he has concerns about how Libra could impact “privacy, money laundering, consumer protection and financial stability.” President Trump and Treasury Secretary Steve Mnuchin are also worried that Libra and other cryptocurrencies might be used to finance illegal activities.
Why it matters: Facebook has given people plenty of reasons not to trust it, so it’s unsurprising that Democrats, Republicans, regulators and even some cryptocurrency experts have reacted skeptically to the project. Expect that backdrop to shape future conversations around cryptocurrencies generally, especially considering how little most people know about the technology.
What’s next: Facebook executive Dave Marcus, who’s heading up the project, is testifying before the Senate Banking Committee today (watch live here) and I’ll be focusing on cryptocurrencies and blockchain in a later issue — stay tuned.
FBI + ICE see you
Article: Drew Harwell / The Washington Post
What happened: Georgetown Law’s Center on Privacy and Technology got their hands on thousands of government records showing that the FBI and ICE have been working with state DMVs, using your driver’s license photos to power an expansive facial recognition surveillance program.
Why it matters: no lawmakers authorized this, none of us agreed to have our information used this way when we applied for licenses, and the program is mostly sweeping up people who have never committed a crime. Facial recognition technology is still highly inaccurate, and especially so for people of color. Additionally, Georgetown Law researcher Clare Garvie called it an “insane breach of trust” for states that allow and encourage undocumented immigrants to obtain licenses to turn around and give their information to ICE.
What’s next: expect more cities and states to face pressure to curtail the use of facial recognition by law enforcement (San Francisco and Somerville, Massachusetts have already banned it) and expect criticism from both sides of the aisle. At the same time, companies developing this technology — like Amazon — and their shareholders will keep trying to sell governments on it because those contracts are often incredibly profitable.
And now, on to Amazon…
Prime Suspect
Amazon is the world’s largest online retailer and second largest company by market cap. Or, for the visual learners (fast-forward to 4:05)…
So, there are a few reasons why it’s important to talk about Amazon:
Scale: Facebook, Twitter and others have shown us that, when companies have hundreds of millions of users, the issues they face and the effects of their decisions are magnified immensely.
Scope: this fascinating investigation from Buzzfeed’s Leticia Miranda shows just how wide-ranging Amazon’s business interests are — beyond e-commerce, it’s involved in:
Apparel (Essentials, Zappos)
Brick-and-mortar stores (Go)
Education (Whispercast)
Energy (Wind Farm Texas)
Finance (Lending, Pay)
Food (Whole Foods)
Gaming (Game Studios)
Hardware (Alexa, Echo, Kindle)
Healthcare (Grand Challenge)
Home goods (Basics)
Logistics (Air, Drones, Robotics)
Media (IMDb, Kindle Publishing, Prime Video, Studios, Twitch)
Professional services (Mechanical Turk)
Surveillance (Rekognition, Ring)
Web services (AWS)
Trust: As of now, people really trust Amazon. The company has made online shopping extremely convenient, cheap and (largely) reliable, while Amazon Web Services has helped make cloud computing mainstream and now powers more than a million sites. That is, a ton of people rely on Amazon for their daily life/business.
Source: Baker Center for Leadership & Governance
Since it’s Prime Day, and because most people think of Amazon as an e-commerce website, I wanted to focus this issue mainly on how that marketplace seeks to facilitate trust between buyers and sellers. However, Amazon’s been in the news a LOT recently, so I’ll touch briefly on a few other key topics as well.
Fake Reviews
If you shop on Amazon, there’s a good chance you’ve used customer reviews to help you evaluate products — and you’re not alone, it’s the second-most trusted type of advertising behind recommendations from friends.
But a recent investigation from the UK-based consumer group Which? found that fake reviews run rampant on the site. To put it in perspective, here’s what it discovered analyzing the first page of results when searching for “headphones”:
100% of products were sold by brands its tech experts had never heard of
71% of products had perfect 5-star customer reviews
87% of reviews came from people who hadn’t even bought the product
5,425 reviews accompanied the top five products alone
Why is Amazon dealing with this problem?
In my welcome issue, I talked about the power of incentives. Here’s how that works for third-party sellers on Amazon:
Amazon Marketplace has a massive, loyal customer base (100+ million Prime subscribers, 300+ million active customers) — sellers, seeing that potential, prioritize reaching customers through Amazon.
However, the competition is stiff (1.7+ million sellers, by one count). That drives sellers to do anything to stand out.
Like Olympic athletes, who face pressure to take steroids or lose out on a competitive advantage by playing fair, sellers face pressure to cheat or risk having their products get lost in the crowd.
Buzzfeed’s Leticia Miranda explored this lucrative black market, where shady consultants help Amazon sellers artificially boost their ratings or improve their placement in search results — often in ways that violate Amazon’s terms of service. One seller, who has used these “black hat” services since 2014, summed up the issue nicely:
Everybody is doing it. People have to survive.
How are people* trying to solve it?
*It’s worth pointing out here that, with any solution, it always comes back to people. Tech companies often say they’re fighting misinformation with artificial intelligence and other technology-driven tools, but those tools don’t design themselves — we still need to analyze and think critically about the people behind them.
Amazon
Amazon has used several strategies in its efforts to stop sellers from gaming the system. First, it targets the reviews directly, using a combination of AI and human investigators who work to identify and take down fake ones.
Amazon also tries to slow the supply of reviews at their source: Facebook, Reddit and other messaging and freelance platforms. It collaborates with those sites to snuff out groups where review opportunities are posted and also has taken legal action against both reviewers and the sellers who employ their services.
Unfortunately, this approach essentially amounts to playing Whac-A-Mole — new reviews and pay-to-play schemes simply pop up too quickly.
So, in 2015, the company introduced Amazon’s Choice, a label indicating “highly rated, well-priced products available to ship immediately.” Though the title might suggest that Amazon employees curated these products, it’s mostly the work of an algorithm. Buzzfeed’s Nicole Nguyen found that many Amazon’s Choice products weren’t good quality, had defects, or had artificially boosted ratings.
Throwing technology at the problem clearly isn’t a silver bullet. The Amazon seller from Miranda’s article I quoted earlier hit it on the head again here:
They’re always going to figure it out. Whatever the machine makes there’s a human to manipulate it.
That’s also why Amazon has been fairly tight-lipped about exactly how it’s fighting fake reviews. It needs to strike a balance between being transparent and revealing too much about its ratings or search results algorithms, lest a malicious seller learn how to better exploit them.
Perhaps recognizing this tension, Amazon has also experimented with different ways to integrate “trusted” human opinions into the shopping experience, but none are perfect:
Vine: an invite-only program where Amazon sends top reviewers free products to review (read this profile of a Viner). But Viners don’t necessarily have the specific knowledge needed to effectively evaluate a product, and just because they know a lot about headphones doesn’t necessarily mean they’re also experts on furniture.
Expert recommendations: Amazon recently began displaying recommendations from independent review sites like BGR, RTINGS and Wirecutter for some product categories. But as with any solution that leans on humans spending lots of time with a single product or review, this approach is really difficult to scale — there are more than 100 million products listed on Amazon.
Lawmakers
Democratic Reps. Frank Pallone and Jan Schakowsky sent a letter to Amazon CEO Jeff Bezos requesting more information on how the company fights fake reviews, verifies products, takes action against sellers who use fake reviews, and whether it makes more money from reviews it claims are verified. They asked for a response by the end of July, but it’s unlikely that Amazon will reveal anything significant there. Additionally, technology moves much faster than Congress, so any legislative proposal could be obsolete by the time it gets passed.
Instead, lawmakers will have to think more holistically about how to disincentivize people from posting — and companies from leveraging — information that misleads consumers and harms ethical sellers.
Regulators
In February, the Federal Trade Commission for the first time found an Amazon seller liable for paying for fake reviews. However, the charges focused more on the seller’s misleading claims about its product (a health supplement) than it did on its use of fake reviews. And given how long it has taken the FTC to bring just this one case, the threat of legal action from regulators is unlikely to discourage sellers — many of them based in China — from utilizing fake reviews.
Third Parties
Perhaps the most useful solution for shoppers in the short-term is to use sites like ReviewMeta and FakeSpot, which analyze listings and identify reviews that appear “unnatural.” A quick look-up of an Amazon product on one of these sites could help consumers suss out products that have weaseled their way into top search results or five-star ratings unfairly. Of course, that adds a step to the shopping process, and people’s aversion to even slight inconveniences may keep these tools from becoming adopted widely enough to affect seller behavior.
Consumers
Shoppers don’t seem to care at all. Market research firm eMarketer predicts that more than half of U.S. households will be Prime members in 2019, up more than 5 million from last year. A few important findings from eMarketer:
89% of customers are more likely to buy from Amazon than other sites
95% of Prime members are likely to keep their membership
94% of all Amazon customers read product reviews at least sometimes
Prices are more important that positive reviews
The takeaway?
Shoppers do rely on product reviews, but they really love Amazon for its low prices and convenience, and trust it to deliver on those promises. So, unless Walmart somehow starts offering a better and cheaper online shopping experience — which given recent news seems highly unlikely — shoppers will probably keep using Amazon.
Ethical sellers frustrated with their competition’s dirty tactics will have to wait for lawmakers to take more aggressive action. Some may try to become less reliant on Amazon by finding other ways to reach consumers, but many simply won’t be able to afford abandoning such a massive market.
As for Amazon, it knows how much power it wields over both shoppers and sellers. So while it’s unrealistic to expect a site that size to have zero fake reviews, Amazon won’t feel pressure to devote additional resources to the problem unless the incentive structure changes significantly.
Finally, it’s important to note the inherent problem in the five-star rating system itself: oversimplification. One person may be an audiophile and have higher standards for headphones than the average consumer. Another may have higher standards for a "five-star” product in general.
While writing this issue, I talked with the founder of ReviewMeta, Tommy Noonan, who framed the problem this way:
You’re trying to quantify an experience that’s entirely subjective and that’s not necessarily quantifiable. There’s this desire to put a number on everything.
Other AMZN topics
Though it’s impossible to cover everything about The Everything Store in one pass, it’s also impossible not to point out some topics that provide important context. My guess is you may already be somewhat familiar with these, but if you’re interested in reading more about any of them, check out the links below!
Worker conditions
Amazon has repeatedly been criticized for the brutal conditions and low wages its warehouse workers endure: 10-hour shifts, 15-20 miles of walking per shift, skipped bathroom breaks, miscarriages and deaths on the job, abusive management practices, 110-degree heat, denied benefits, and unlivable wages. This is all while the company raked in record profits last quarter, took at least $2.6 billion in taxpayer money in the form of subsidies, and paid zero dollars in federal income taxes.
Amazon Prime Day comes with protests by warehouse workers, pilots and engineers (Kate Gibson / CBS)
Amazon Gets Tax Breaks While Its Employees Rely on Food Stamps, New Data Shows (H. Claire Brown / The Intercept)
Alexa and data privacy
Amazon says that it has sold more than 100 million Alexa devices despite the fact that more than 70% of Americans are concerned about privacy when it comes to smart speakers. But that’s also assuming those polled actually understand how Amazon listens to their conversations or uses their data — the evidence shows they don’t. People may ultimately be okay exchanging privacy for something valuable in return, but it’s important that they have accurate information about what that tradeoff is.
Lawsuits claim Amazon's Alexa records kids without their consent (Ben Fox Rubin / CNET)
Amazon and Google are listening to your voice recordings. Here's what we know about that (Ry Crist / CNET)
People say they care about privacy but they continue to buy devices that can spy on them (Rani Molla / Recode)
Facial recognition technology
Amazon employees, immigrant groups and others have staged multiple protests in the past week, taking aim at — among other issues — the company’s ties to law enforcement, specifically with regards to its facial recognition technology, Rekognition. Critics of the technology worry that it could enable human and civil rights violations, especially given the recent news about the FBI and ICE.
On Prime Day, Amazon workers and immigrants’ rights organizations are protesting (Megan Rose Dickey / TechCrunch)
With No Laws To Guide It, Here's How Orlando Is Using Amazon's Facial Recognition Technology (Davey Alba / Buzzfeed)
San Francisco bans city use of facial recognition surveillance technology (Trisha Thadani / San Francisco Chronicle)
Antitrust and monopoly power
One theme underlying each of these issues is the question of whether we’ve entrusted Amazon with too much power (either knowingly/intentionally or not). As Amazon’s business empire expands, it continues to consolidate both economic and political power. Though it often downplays this by focusing on numbers like its 4% share of total retail sales versus its 38% share of online sales, regulators and lawmakers are growing increasingly concerned about Amazon’s potentially anticompetitive business practices. That said, there are complicated technical, economic, legal, social, political and other issues involved, and coming up with workable solutions will be easier said than done.
Amazon’s Antitrust Paradox (Lina Khan / Yale Law Journal)
The push to break up Big Tech, explained (Matthew Yglesias / Recode)
What to expect from tomorrow’s antitrust hearing featuring big tech (Jonathan Schieber / TechCrunch)
Next week + feedback
Hopefully, this rundown gave you a better sense of the different ways we trust Amazon and enables you to think more critically moving forward about when and why you should — or shouldn’t — trust it. Next week, I’ll be taking a look at how Uber and Lyft use their own five-star rating systems to signal trustworthiness.
Any feedback on this issue? Click here:
Haven’t subscribed yet? Get Trusty in your inbox each week!