Specifications for an alternative/new DHT

Branching off my noob question series, assume I’m also asking if I should.
Sounds like the current DHT may not be performant enough to act in the roles I desire. Through the extensible nature of libp2p, it’s my guess that there’s room for more than one DHT. My task is that of designing, building and implementing a database that would support most or some of the things I’ve outlined in the noob series. The goal of this thread is to outline/set targets and expectations. This is well outside my current skill level to complete. I’m confused that I don’t have this already, some projects are close but end up missing the mark in crucial ways.

  1. Retention of weeks, or at least > a few days. Long enough for a node that connects twice a week to see everything.
  2. Inserts on the order of a few per second for sets of 3 to 4 minutes, then a to be determined pause. No reason, it’s just that DDR seems like a decent target.
  3. Size to be determined.
  4. Anti-spam/proof of stake?
  5. WoT liking/disliking data should be used to keep/discard inserts.
  6. WoT profiles should be higher than decisions on inserts.
  7. Inserts can have TTL.

As I’ve tried to describe, my goal is to have games use network resources that are controlled by the players. The main reason is longevity, to protect against things like D2LoD BattleNet servers being decommissioned.
With an eye towards extensibility, unlike the perception GnuNet has.
To eventually create an open FreeNet replacement, though this isn’t meant to be a file warehouse… Unless it can be and still meet the insert and retention goals, I assume these are mutually exclusive.