ODDOSCOPE ORIGINAL / STORY / 3 MIN
A Feed Is a Choice
Three posts, one top slot, and a surprisingly important definition of “better.”
From The Internet & Your BrainYour new job: choose the first post
Imagine a tiny, fictional social app called Window. This morning it has exactly three posts: a joke from your friend, a careful explanation of a local bus change, and a photograph from someone you have never met. You have been promoted to head of the feed. Congratulations. There is one slot at the top, and absolutely no welcome hamper. Which post goes first? “The best one” sounds helpful until you have to write the instruction. Best for a quick laugh? Best for getting to work? Best for discovering someone new? The posts have not changed. Your purpose has.
A prediction needs a target
Suppose Window rewards the post most likely to receive another click. The joke looks promising. Now change the target to helping a reader understand something useful. The bus explanation becomes a stronger candidate. Change it again to introducing an unfamiliar perspective, and the photograph has a case. These are invented editorial choices, not estimates of a real platform's ranking. Actual systems are more complicated than our three-post desk. In its own public description, YouTube says recommendations consider viewing behavior and signals including likes, dislikes, subscriptions, feedback and satisfaction surveys. That is the platform's explanation of its system, not an independent audit or a complete recipe we can reproduce.
What the scoreboard leaves outside
Even a perfectly calculated score cannot settle whether the target was a good one. A click could mean interest, confusion or an accidental tap. It does not come with a little certificate saying “this was worthwhile.” A useful instruction therefore needs both a goal and a way to notice when the convenient measurement stops serving it. For Window, you might count clicks but also ask whether readers understood the bus change. You might reserve a place for unfamiliar contributors. Each extra rule introduces a tradeoff: limited space, uncertain predictions and people who want different things. Calling the resulting order automatic does not remove those decisions.
Try a different brief
Write two instructions for our imaginary feed: one for a reader with thirty seconds, another for a reader hoping to encounter a new idea. Choose the leading post under each instruction. Then name one person your choice might serve poorly. You do not need a universal winner. You need to be able to explain what the order is trying to achieve. In You Control the Algorithm, six small decisions let you practice making that explanation. The scores belong to an illustrative game. The question you can take back to your own feed is simpler: what seems to be rewarded here, and what would I rather reward?
Sources & editorial notes
Rights: Original Oddoscope text and code-authored SVG illustration. External references are linked for attribution, not reproduced or licensed by the presence of a link.