Do you want to elaborate more on how politeness cant be explained by gricean maximes?

The Gricean maxims only handle the informative part of a conversation; they don't handle, for example, the emotional impact of the utterance on the hearer, or the social impact on the speaker. As such, in situations where politeness is a concern, you'll see people consistently violating those maxims.

I'll give you an example. Suppose two people in a room: Alice and Bob. Alice has a lot of cake, she's eating some, and Bob is craving cake.

If Bob were to ask Alice for some cake, Bob could simply say "gimme cake". It fits the four maxims to the letter - and yet typically people don't do this, they request things through convoluted ways, like "You wouldn't mind sharing some cake with me, would you?" (violating the maxim of manner), or even "You know, I was in a rush today, so I had no breakfast..." (implying "I'm hungry", and violating the maxims of quantity and relation).

To handle why Bob would do this, you need to backseat Grice's Logic for a moment and use another framework - such as Brown and Levinson's politeness theory, it explains stuff like this really well.

This is probably obvious for you (and for me), and yet you still see some pragmaticists shoehorning everything into Grice's logic. Or some doing the exact opposite and shoehorning it into Austin's speech acts, or B&L Politeness Theory, etc. It sounds a lot like "I got a hammer, so everything must be a nail".

To the latter point: My biggest gripe with linguistics is the tendency to boil everything down to a simple system.

Yes, yes, and yes. You can see Language (as human faculty) as a single system but, if you do so, any accurate representation of that system is so big that it's completely useless, like a map as large as the territory.

That's already a tendency in Linguistics in general, but in the case of the generativists it's their explicit goal.