From the wiki:

Total Food Consumption Formula

Food consumption = 0.25 * P^2 + 6 * P

P = population count

Taking the derivative shows that each new pop increases the consumption by roughly 0.5*P + 6 food. We've had lots of threads on the weird edge cases this creates (merging cities which immediately starve) but even when it's working as intended... why? Given that religion and influence spheres (which would presumably play off population) basically don't exist, the only purposes pop serve are for creating units and working in pop slots. Pop costs for units are effectively divorced from food yields because of the variable consumption. There's a bit of math that causes the pop to grow back slightly faster if the city was big to start, but either way unit pop cost is nearly free because the city grows back to equilibrium so we ignore that (and the pop buyout, for multiple reasons). So the only real use, pop slots, is just wonky.

If you have 2 pop in your first city, they'll consume 13 food per turn and putting them in un-upgraded farmer slots gives... 12 food. 14 with Tribe's Legacy. Switch that to a different yield and food basically becomes a wildcard resource with roughly 1:1 exchange rate. That's pretty good right? The flexibility is a big deal, especially early on, let's see what happens when it grows some more...

4 pop consume 28 food and give 28 yields. Still good.

8 pop consume 58 food and give 56 yields. Flexibility might be dipping a bit if slots haven't been built, but still good.

16 pop consume 160 food and give 112 yields. This is where it gets rough. Sure there are some effects in the game that increase pop scaling, but they're hardly common by the time you reach 16 pop. And even then you're committing to research/build something just to slightly linearly increase the value of the wildcard resource, while trying to keep up with exponential scaling. But ok, let's say we get a few of these effects...

32 pop consume 448 food. At this point the exponential term is responsible for 256 of that so the growth is almost entirely exponential. Ok, but a wildcard resource is still useful right, even if it's easy to build for any resource you want at this point... you'd need 14 yields per pop to get that 1:1 return. Maybe you consider that a worthwhile investment, let's keep going...

64 pop consume 1408 food. Yeah. We can all see where this is going. Even if you got the yields to 22 per pop to reach 1:1 again... is that a win? Not to mention, what are you working towards? In a game of exponential snowballing, an uphill battle to sustain linear growth isn't a winning strategy. The wildcard resource gives progressively worse returns until it borders on useless. Food is not a win condition in the same way that the other yields (especially production, and money if the buyout costs weren't borked) are. It's not useless, but the amount you get without actively focusing on it is plenty for most purposes so there's little reason to actively pursue it.

This is nothing new of course. EL was just slightly better. The idea of depicting what's supposed to be an exponentially growing population (which makes sense since it's depicting populations from the bronze age to now) on a linear scale (which doesn't make sense except for game mechanics) is fundamentally wonky. It's flown under the radar for the most part but HK (and ES2 to a lesser extent) exposes the contradictions in that system by letting us migrate pops (highlighting the linear/exponential representation inconsistency). That said, the stinger is the existence of food cultures. The implication that food actually matters enough to pick a food culture, much less that "holy crap I have so much food this is amazing!" is something someone could ever say in this game, is laughable. The economies in EL and ES2 were always pretty wonky but it worked because the games never suggested that a given resource was worth focusing on. Players found a meta yeah, but it was never blatantly at odds with what the game was visibly pushing. HK cultural affinities isolate pieces of the economic system and people are catching on to how broken it is as a result.

Is this fixable? The way I'd fix it is waaaay out of scope for the devs, probably even for mods (abandon the idea of pops altogether, represent size by districts instead and have districts consume food, food surplus becomes a percentage like stability that throttles the number of districts you can sustain in place of that stupid production cost scaling). Bluntly forcing cities to require population somehow could do it, I've used a mod that adds a malus if the district/pop ratio gets too out of line. Some new mechanic that makes pop genuinely important enough to fight the uphill food treadmill could do it. Realistically though, HK will hopefully be the push the 4X genre needs to move away from Schrodinger's Linear/Exponential Population.