Amagooglezon

Amazon has a whim machine bolted onto their ecommerce system. The recommendation engine is a combination of practical — “Other people who bought X also bought Y” — and bemusingly enthusiastic: “You clicked on X before so I bet you’d really like ALSDJFLKSAJF too! Wow, look at all those letters! Notice how they’re in the same alphabet as X? Pretty impressive, huh?” It’s bad at nuance but it’s good at throwing out options. There are so many options for it to scan and suggest.

This stock photo amuses me. Image via Robbert Noordzij.

Businesses need to solve hard problems in order to be successful. Shopping on Amazon is cheap and convenient and they have a vast array of goods. Selling things cheaply without collapsing is a hard problem, and so is convenience, and so is being stocked with lots of products. Amazon conquered all three challenges. Now the benefits feed into each other. Customers love the cheapness and convenience, so sellers must stock their storefronts. Sellers are much easier to aggregate than customers, so once you figure out the customer bit, you’re golden.

Superstar internet businesses — and I guess most high-value companies in general — are all about positive feedback loops. Circular incentives that channel energy from initial success to intermediate success to dominance.

When you search for something on Google, you make Google better by feeding data into their algorithm, and your presence incentivizes both websites and advertisers to cater to this particular search engine. You come back because the algorithm is so good at presenting the information you want. Websites worry about SEO and advertisers drop $$$$$ because that excellent algorithm keeps pulling you back. The incentive structure is great for Google. That ingenious feedback loop made them dominant and it keeps them dominant.

Fast forward to 2037 when we’re surfing Amagooglezon (or whatever supplants them) with our heads swimming in VR buckets. We’ll bounce from product to product, purchasing and appraising and reviewing and returning and diving into on-demand experiences. I wonder how recommendation systems will work then — maybe they’ll have personalities. Maybe we’ll fall in love with them. Maybe we’ll hate them. Maybe our wallets will be managed by AI assistants and none of this will matter.