Pay to Play
Ads
Much is made of Google and Facebook’s dominance of the digital ad space. Between the two they control the vast majority of ads served online - Google has a near total monopoly on search results, and runs the largest display ad exchange. Facebook and Google combined bring in more than half all digital ad revenues in the country. In a distant third with thirteen percent is Amazon.
You might be thinking - Amazon? Why is Amazon taking in so much in advertising revenue? The answer is - not content to take up to 50% off the top of their marketplace sellers, Amazon is now operating a multibillion dollar payola* scheme to force the same sellers to pay for ads.
Amazon insists it is merely an e-commerce platform, which gives it cover to pick winners and losers for cash. Rather than simply listing the top rated or lowest priced products for a wide category like ‘vacuum cleaners’, Amazon charges merchants money to appear above the regular product listings. Subtly marked ‘Sponsored’ products pepper the real results.
The most insidious ads I’ve come across recently on Amazon appear in their mobile app - author searches. Ironically, Amazon has become a worse and worse place to purchase books, and allows rival authors and publishers to snipe searches for author names. A search for Hugo-winning author Arkady Martine gave me two sponsored results from different authors, one that happens to be a part of Amazon’s paid Kindle Unlimited subscription. There’s also a sponsored top scroll with another author’s work, unrelated to my search. Martine’s work doesn’t appear at all above the fold.
How demoralizing must it be to win one of the most prestigious awards in your field and have the Amazon search results for your own name be full of other people’s work? Here’s another celebrated sci-fi author with multiple TV and movie deals for her work:
The problem is authors and their publishers have to be on Amazon, because the company has leveraged its monopoly to force out competitors. If they decide not to participate in the platform’s pay-to-play model, their books will be listed beneath anyone who will.
Speaking of payola schemes, let’s talk Uber. In a somewhat surprising announcement, the company reported its ad business is growing quickly, driving revenue for a company not known for making money at anything. How is it doing this?
Uber sees more opportunity for growth given that only 25% of the companies selling products through Uber Eats are buying ads on the platform…
Just like Amazon or Spotify, restaurants who want top listing on Uber Eats will be expected to pay for it to compete with their rivals. It’s not bad enough the company charges huge fees to restaurants, they now have to pay for the opportunity to fight over those scraps.
These advertising revenue numbers may placate Wall Street, or present the illusion of growth, but to what end? If Uber or Amazon expects every one of their advertisers to pay for search results, their apps will become a sea of poorly-vetted sponsored content just like Google, and consumers will be even worse off for it. We have come to expect a certain level of advertising online, but too much makes us miserable. Not to mention, payola advertising undermines what little consumer trust remains - people expect searches to yield credible results, on a web store or a food app or the world’s biggest search engine. The problem is, companies don’t care about accuracy of results anymore, they’re only focused on how to more creatively cram more and more ads down our throats and, failing that, are turning to the merchants on their platforms to squeeze the remaining blood from that stone.
*Note: Payola is commonly used to describe bribing radio DJs to play your songs, but it’s a great word and I support its application to other self-dealing industries, like platform ads.
Also Note: if you’re looking for a way to see less ads, try uBlock. It works on Google and Amazon - but only on desktop. Mobile users are still out of luck.
AI
The AI media cycle reads something like this:
A tech company releases an AI bot capable of doing one or two clever things.
People find fun and interesting ways to use the bot, break the bot, or make it say slurs.
The tech press writes breathless columns about the power of AI.
Pundits, academics, and bloggers imagine the ways this new AI is going to replace humans.
The tech company takes the bot down to ‘optimize’ it to stop breaking and/or saying slurs.
Everyone forgets about AI for a few months.
It feels like we repeat this cycle a few times a year, no? We’ve talked before about the various failed chatbot rollouts that were supposed to revolutionize AI, and whenever a new bot is launched we act as if those past failures happened to different people, on a different planet.
ChatGPT has more staying power because its creators spent years paying low wage African workers to fine tune its responses to be coherent and slur-free. The most important bit of any AI/chatbot/ML model is the database it draws its information from. OpenAI spent years scraping, filtering, and - most critically - hand-cleaning data for its ChatGPT bot, which helps it have convincingly human conversations.
I bring up the bot’s database because when Microsoft - OpenAI’s biggest investor - took the same AI model and applied it to their data, weird shit happened:
Microsoft’s newly revamped Bing search engine can write recipes and songs and quickly explain just about anything it can find on the internet.
But if you cross its artificially intelligent chatbot, it might also insult your looks, threaten your reputation or compare you to Adolf Hitler.
I keep pointing out the problem with attempting to train a chatbot on a dataset as large and dirty as a search engine index. The only two companies with the resources to index most of the Internet are Google and Microsoft, with the former spending far more time manually cleaning those listings. Is it any surprise that bots plugged into these massive datasets full of junk information and shady SEO websites are ornery, or downright threatening?
Speaking of humans cleaning up software messes, two stories made news last week and I haven’t yet seen anyone connect the two. Tesla was forced to ‘recall’ (unsure what this actually means!) it’s Full Self Driving Beta software, because it was programmed to run lights and make dangerous driving maneuvers, and also because it is not actually self driving, though they haven’t yet made Tesla rename it. The company was also hit with an NLRB complaint saying it illegally fired employees at its Buffalo factory who are trying to unionize. Multiple publications cite over eight hundred employees at the factory, many or most of whom work on processing the data to train the FSD software. That is a lot of people! It is also a reminder that when or if AI bots do ‘replace’ some human functions, they can’t operate without a lot of humans ingesting, interpreting, and providing clean data results for the bot to operate from. We’re seeing now what happens when that job isn’t done properly - we get snarky bots and cars that can’t detect fire trucks.
Anyhow, one task AI might actually be suited to is radio DJing, which involves inserting a few dry quips in between songs. Spotify’s got an AI DJ, which it insists is going to help listeners learn more about artists, which is fine, but if Spotify wanted to actually help artists they’d pay them more than one one thousandth of a penny per stream.
Elon Musk
One thing that often happens in the finance press is they equate a valuable public company with a valuable private company, or imply the two are interchangeable. For example, Elon Musk recently won a shareholder lawsuit that accused him of making false claims in his infamous ‘funding secured’ tweet saying he was going to take Tesla private. Musk claims he was serious about the deal, though in hindsight it was very, very good for Elon that he did not take Tesla private, because the stock went on a tear after that and his net worth ballooned.
What if Elon had taken Tesla private, though? All the reporting I have seen about the incident discusses whether he could have secured the $60 billion dollars or so he’d have needed to buy the outstanding Tesla shares. Practically nothing is said about what would have happened afterwards. Let’s say Elon is now King Volt or whatever at Tesla, the company is private and he can run it however he pleases - I’m not sure that would look any different from how he runs it as a public company - but whatever percentage of Private Tesla he owned would be…not worth anywhere near as much as his shares in Public Tesla.
In 2018, Elon may have been optimistic about Tesla’s future - the company was about to nearly double its revenue YoY:
This is positive - setting aside the fact the company was ‘profitable’ due to emission credit sales via government subsidies - but if Private Tesla had made $21 billion dollars while taking on something like $30 billion in debt and investments (Musk couldn’t have sold his Tesla stock to buy Telsa), what happens next? The company made a lot of revenue by scaling up its operations, but its net profit was actually declining as it spent money to expand. Public Tesla didn’t give a shit because its stock price kept going up and up, buoyed by Musk’s public persona as a Wall Street darling.
If Private Tesla had taken on, I don’t know, let’s say ten billion dollars in debt to get the deal done, it might owe a billion dollars in interest payments, and the investors who chipped in the other thirty or forty billion might be asking Musk why he was willing to burn billions more a year when they’d be on the hook for the difference. Private Tesla would have to either seek new investors - if existing ones agreed to share dilution - or take on more debt to expand, since the company wasn’t bringing in the profits to justify exponential expansion.
I am bringing all this up because we have a real life example of what happens at an unprofitable company run by Elon Musk with billions in debt: Twitter! Musk’s erratic whipsawing between bankruptcy threats, firing most of his staff, refusing to pay rent and vendors, and rolling out features no one asked for while advertisers flee the platform gives us a look into what Private Tesla might have faced if Musk had gotten his wish. In 2018, he was not ‘worth’ a hundred billion dollars or whatever, because Tesla’s stock price had not gone up fifteen hundred percent, and so he wouldn’t have had the leeway to continue to burn billions in new investments at his car company.
The finance types might say ‘but the company would still be worth billions! he could take it back public! what about SpaceX!’ Sure, Private Tesla and Private SpaceX are probably worth a lot of money, maybe even more than Musk and his investors would have paid for them, but the more debt the companies took on in lieu of access to the cash firehose of public markets, the less investors would be willing to pay for them.
The point, as always, is that ‘growth’ companies - companies that reinvest their meager profits or spend way beyond their private borrowing power - are built on brittle foundations propped up by market sentiment, founder charisma, or some combination of factors that leads stock analysts and bankers to collectively agree that Tesla, a company that made 350,000 cars in 2019 at a break even despite billions in tax credits was somehow worth more than all its competitors combined because some day it might be worth ten times what it was.
Annoyingly, they were…sort of right? Tesla has increased its market value nine times since Musk nearly made the worst business decision of his life. Since cresting at an absurd trillion-dollar valuation, the company sunk as low as a mere $350 billion before nearly redoubling this year for…reasons? Like Musk’s erratic behavior, the value of his car company seems only loosely linked to its financial performance, business plan, or production forecasts.
Late last year, Musk frantically cut prices on Teslas to prop up the year’s flagging sales. Numbers in the US, Europe, and China indicate Tesla might be reaching market saturation - here in the US, the F-150 is the top selling car, and the idiotic Cybertruck isn’t going to threaten that market. So, is Tesla worth $650 billion dollars if its sales level off or decline, and Musk’s continued obsession with Twitter causes his investors to lose confidence? Probably not, because that number is a fantasy, created by Wall Street to enrich itself and prop up a highly lucrative if personally awful CEO. Finance press will continue to report on Twitter as a cataclysmic disaster of bad management, while insisting the same guy is running Tesla and SpaceX like an unadulterated genius.
Musk is very good at taking lots of investor money, promising a bunch of stuff that never comes true, and riding stock prices upward. It turns out he really, really sucks at running a private company that depends on things like advertisers or developing good software to make money. Somehow we’re going to talk about Genius Musk and Failure Musk simultaneously, forever.
Food Labeling
You know the vibes:
General Mills, Kellogg’s, and the rest of the country’s cereal makers are mad at the FDA. So are the packaged food companies, the pasta industry, and the pickle lobby (yes, it exists).
The companies behind America’s favorite culinary indulgences are worried their products wouldn’t be considered “healthy” under a recent Food and Drug Administration proposal — and they’re urging regulators to reconsider.
Pickle lobby?! Of course it exists. So what are the food companies all het up about?
The FDA put out the guidelines at issue back in September, arguing that to be marketed as “healthy,” foods would have to include a certain amount of key nutritious ingredients, like fruits and vegetables, and have little added sugar, sodium, and saturated fat. The agency’s proposal would not ban unhealthy foods; those that don’t meet the FDA’s standard simply couldn’t be labeled as healthy.
This seems reasonable! But, unfortunately, processed food companies have built brands around this and, well:
Even the maker of the frozen-aisle favorite, Healthy Choice, says it couldn’t follow the FDA’s new guidelines “without alienating consumers.”
The food lobby is very upset about this, because obviously they are. What’s funny is the FDA - who we’ve roundly criticized around here - didn’t even set a particularly high bar for healthy, and food companies still can’t meet it:
A frozen Salmon meal with green beans and rice can’t have more than 2.5 grams per serving of added sugar, more than 690 milligrams of sodium, and more than 4 grams of saturated fat to be considered healthy, according to FDA’s website. (Healthy Choice’s Barbecue Seasoned Steak Dinner has 16 grams of added sugar, though it meets the FDA’s criteria for both sodium and saturated fat.)
We’ve talked about all the damn sugar and sodium in American food and while I don’t think consumers focus in on the ‘healthy’ part of the labeling at this point, maybe they do? What are they going to rename Healthy Choice? I can’t wait to read the lawsuits.
Short Cons
Gizmodo - “Looking for lists of people with depression, anxiety, bipolar disorder, PTSD, or OCD? No problem. There are lots of companies who would love to sell it to you.”
POLITICO - “The Labor Department’s internal watchdog identified nearly $30 billion more in pandemic unemployment benefits that were wrongfully sent out than previously estimated…”
ProPublica - “The state of Louisiana is dropping thousands of lawsuits against homeowners who received grants to elevate their homes after hurricanes Katrina and Rita in 2005 but used the money to make repairs instead.”
Bloomberg - “Companies avoid paying about $4 billion in overtime wages by inventing dubious titles for US employees such as “director of first impressions” and “lead shower door installer,” according to new research on a common practice that skirts federal labor law.”
Know someone thinking of starting a healthy brand of processed foods? Send them this newsletter!