1. you forgot 8-12 from caravan, city states, pantheons... the AI proved me you can have fairly well populated cities in low food locations. 2. i've hardly ever had a reasonable position without floodplains or oasis or any sort of food tiles. that's very dependent on map script and map size. (biggest map i play is tectonic standard, because i dislike slower speeds and want domination victories to be possible.) 3. of course you need to have insanely good reason to settle in said location. just imagine you had a crazy good strategic position, with an insane amount of strategic and luxury ressource hills that would be crucial to the monopolies. but no . would you settle it? you would, and you would do everything to make the city thrive. but you don't always have such a crazy good location that you chose to settle. the reason is, in such areas, like in real life, has such a high value that you need to have a huge payoff to profit from settling it. it's a trade off. you're paying a caravan slot, , and maintenance for buildings just to make up for the fact that there's low food in the area. but it CAN be worth it. and if so, every yield in the city is insanely valueable, translating in a huge value factor for in the TileValue formula that i mentioned earlier. the factor drops as the city has more and more food income, to the point where, if you had a city with only, exclusivly food tiles, the value factor of would become 0 or close to. reality is somewhere in between. also, i'm not saying i'm opposed to buffing flat deserts (i'm neutral), but i'd rather see tundra buffed. it's a much more common starting location and consistently bad (not even features make up for it), especially if you don't get the pantheon, as opposed to deserts which have 2 pantheons, flood plains and oasises, petra and their hills don't suck. and they're not close to snow.