November 30, 2018
On mobile? use this version and hit full screen
EDIT: Thank you for all the responses and feedback! Here are some of my favorite insights:
From Rod Frey:
One datapoint: Youtube is my primary content delivery platform right now, I subscribe to about 30 creators and spend about 3-5 hours a week watching them. I don’t think I’ve ever discovered a creator through the recommendation algorithm. Instead, all new creators have been added because of recommendations from people I’m already watching – shoutouts. I think that’s actually pretty common: I’ll often be looking at the comments and see a huge string of “Great stuff! XXXX sent me!” type comments.
So it could be that communities naturally band together to disintermediate the algorithms. That probably doesn’t work for creators who are in competition with each other, which probably happens at the top financial levels of Youtube, but it seems to be a significant factor for special-interest communities. It’s also more immediately actionable than more aggressive disintermediation like building an app or collecting emails somehow – people can just start promoting each other. (Obviously this is only relevant to media creators, it’d be harder if you’re making and selling sassy socks on Etsy).
From Misha Berger:
…recommendation algorithms are not the only way content is discovered, even if it’s a major part of it. There’s also word of mouth, seeing the objects in everyday use, endorsement by influencers (not all of whose content is delivered algorithmically), media (not all of whose content is delivered algorithmically), physical ads (billboards, public transportation, etc), and so on.
A recommendation from Hannah Marcus:
A personal experience from the e-commerce space dealing with Amazon, from Alex Furmansky:
I like how you showed this as an AI issue… but I’d argue that it’s a broader platform issue that existed before AI got involved. Etsy is a good example of this… search for “custom ring” there… tons of artists… all with one boss… Etsy. Good luck firing Etsy, as you said. Instead, have fun paying Etsy for promoted listings and all there other “upgrades to help artists”… none of which would be necessary if you weren’t competing against dozens of similar artists from around the world. Doesn’t matter if Etsy has AI or not… just the fact that they are a network effects marketplace business gives them ultimate boss power.
Same with Amazon: there are dozens of sellers fighting to be the one-click buy button (and dozens of products fighting in seemingly perfect competition for any search term). All of these sellers looking to make a quick buck have one boss: Amazon. And Amazon absolutely fleeces them. I was invited to acquire (more accurately: save from bankruptcy) a toy retailer whose primary sales channel was Amazon. The owner showed me over $1MM in sales with normal manufacturing margins… how was he bleeding money? Turns out Amazon was charging him seller fees, stock fees, fulfillment fees, advertising fees, and… best of all… interest on the loans Amazon gave him so he could pay for the advertising fees in the first place! Below is a sampling of emails Amazon sends to merchants… this is within just a 60 day period… notice how heavily they push advertising on Amazon? The boss wants all of the “coworkers” fighting against each other and bribing the boss so the boss picks them as a favorite for that day.
To extend your metaphor: Amazon is not just a retailer… and it’s beyond a boss… it is more like a king that makes rules and then taxes the serfs to oblivion.
From Karan Singh:
I liked the advice you offer to be “antifragile” and enjoy the upside.
Something that I don’t quite agree with is the personification of algorithms – maybe it’s a oversimplification for purpose of presentation but it stops short of asking some meaningful questions imo: why do humans design such toxic algorithms? Do we not realize the toxicity? Do we overlook it to drive business value over everything else? Why are the most popular discovery platforms supported primarily by ad revenue?