The point is not, that it would be easy to auto-ban some bots or multiaccounts. I never said anything about a "no-false-negative"-problem. The point is false positives. And your pseudo-code would result in a lot:
- A player can't even send 80% of his resource production to anyone, pushing protection limits this depending on the connection between the accounts (same pc/ip, sitter relation, no connection)
- If a player sends the pushing protection limit it can have many reasons, for instance kingdom-intern temporary pushing e.g. to compensate recent off losses (which is fine and no multiaccounting), an agreement between farmer and farm (which is fine aswell), ...
- If a player is farmed beyond good and evil he maybe is just a farm and no multiaccount?? There are at least 15 players in my current game world who lose 80%+ of their production to me without them being my multis
- A player not reading messages could be using the app and not see/read them there
- A player insta-reading all messages might just have left the chat windows opened ...
- If the game requires me to solve some fucking captchas every day I would insta-quit it, because I'm retarded and need like 10+ tries at every single one, those fuckers are just annoying. Or imagine making an offplan and not be able to time attacks due to these fuckers
- If a newbie doesn't know how to behave when being farmed, barracks etc. are being destroyed (which isn't that unusual) and are farmed 24/7 you just can't do too much
- How to you tell a program to differenciate between "random letters" and normal village names. For instance, once I named my villages after a chess game (like 01. e2-e4 e7-e5, 02. Ng8-f3 Nb8-c6, ...), I know one who named it after chemical elements (which involve a lot of numbers and letters), and so on.
You just can't come up with a good, relieable auto-ban system that has a 0% false positive rate. As I said, from a programmer's view it's just too easy to hide MA/Botusing. Sure, you can catch some idiots that are MAing, but that's the case right now with manual-only checks aswell. So no improvement there. And you can code detection versus a specific bot, but those will probably change their behaviour often enough.
The point is, to relieably tell apart human from bot and MA behaviour, you need to think and analyse a lot. This is likely a bachelor thesis and more worth of analysing, to be honest, I can understand that they don't implement it. AND - I wouldn't want any system that can even think of banning a non-bot / non-ma mistakenly. Why? Obviously the devs won't tell the detection mechanisms, so the MHs & CMs have to rely on what the system sais. So if it's a false positive ... he's just fucked.
There are reasons, why this isn't a thing (yet?).