New idea for an intelligent blocksize increase

[balu2]

I came up with an intelligent solution:

just make an update which scans the last 2000 blocks mined for their size and if those blocks were on average more than 80% or 90% full the network could automatically produce blocks 0.1 MB bigger. So whenever for a long time the blocks are full or near full an automatic increase by a very small step like 0.1 MB or less could be done by the network. Ideally the increase is a little slower than the build-up of fees to keep the chain secure.

I think this proposal from a non-coder is a lot better than that bullshit Gavin produces. Opinions?

Code devs can feel free to use my idea like it was their own.

Increasing blocks from 1MB to 8MB in a single step without market-demand for this blockspace or consensus is the idea of a moron, clearly.

If blocks should be bigger, they should be bigger because of demand and not because some moron says so.

[Delek]

If a malicious and powerful miner decides full all blocks, it will end up having 8mb blocks after aprox. 3 years. Seems rare, maybe we need more speed on that.
(2000 blocks are aprox 2 weeks, 1 + (0.1/14)*365*3 = 7.82mb)

To stop block size from increase without a reason, what about a fee per kb in blocks for miners?
And what if we need MORE than 0.1mb of increment?, maybe adding a % of block increase would be better.