US dog treat and chew market revenue

It appears that somewhere around 2010, the universe decided that the number of artificial intelligence researchers publishing papers on arXiv should move in perfect lockstep with the amount Americans spend on treats for their dogs, as though both were being controlled by some distant cosmic bureaucrat with a peculiar sense of proportion. One might expect these two phenomena to have roughly the same relationship as a cheese grater and a submarine, yet here they are, climbing together with the determination of old friends on a steep hill. The correlation is 0.988, which is the sort of number that makes statisticians uncomfortable in ways they struggle to articulate at parties.

The explanation, it turns out, is probably not that machine learning algorithms have become sentient and are demanding premium peanut butter biscuits as tribute. What's more likely is that both trends ride atop the same wave of underlying economic expansion and technological enthusiasm throughout the 2010s—a period when Americans generally had more disposable income, pets were increasingly treated as family members worthy of indulgence, and academic publishing in computer science experienced a genuine renaissance funded by tech industry money and institutional momentum. The AI paper count grew from roughly 200 per month in 2010 to over 1,500 by 2022, while the dog treat market expanded from about $2 billion to $4.5 billion, both moving on the current of the same rising tide of wealth, interest, and the peculiarly modern conviction that if something exists, we should quantify it obsessively.

Which brings us to the rather humbling realisation that we humans are pattern-seeking creatures living in a world where patterns genuinely exist—the trick is knowing which ones matter and which ones are merely two unrelated boats rising with the same economic tide. Perhaps the real lesson here is not about dogs or artificial intelligence at all, but about how a correlation of 0.988 can tell you almost nothing while simultaneously telling you something oddly true about the 2010s. The universe is under no obligation to make sense.