Just adding our two cents here. Some of these concepts have already been discussed back since the Ethereum Frontier days. So we just want to reiterate some of these points plus our view on untransferrable NFTs aka "soulbound tokens" which in reality relates to identity in our view.

(2) voting power calculation which favors small holders (for example less than 100 DOT: x50 - 101-500: X20 -500-1000: x10…). But this bonus could be used only with verified account to avoid bots.

This one has been already tried on Kusama when one of the proposers didn't get his way with the treasury. He proposed a higher weight for small holders. The problem with this is that it's gameable. In this way, a whale could simply break his holdings into bins that have the highest weight and have even more power than regular small holders. Verified accounts don't help much either since you can create a different identity for each one and still game the system. It's not a complicated setup either.

(3) quadratic voting: For example Token holders can used their voting power only once or twice during a decision period (28 days). Token holders wil vote for the most important proposals for them. Quadratic voting put another game theory in Polkadot governance.

FYI Quadratic voting is already implemented since day 1, expressed by conviction voting. As for limited conviction. What would happen when 2 or 3 very important referendums appear within the same decision period which the token holder would like to vote because he has strong opinions about. The thing with limiting voting conviction is that it makes it harder to delegate as you'll also have to rely on the conviction of the delegate something that the token holder might not agree with. Moreover, scam or grifters will take advantage of this, so they can add  their referendums in moments of several controversial referenudms so that high conviction votes are spent on other, more important referendums.

(1) on chain reputation system with Soulbound Tokens

This is an interesting one and one that shows promise and related ideas are already in research. But it should be implemented and tested heavily first within other DAOs where reputation matter. Though the idea would be more fitting within dapps that require more reputation than blockchain governance, in our opinion. We saw the discussion by Bruno Skvorc on X about linking reputation to NFTs which makes sense in some setups. However, they don't require to be unmovable. These in fact could be linked to on-chain identities and moved within the same identity whenerever needed, after all Polkadot and others don't have the same limitations of EVM chains. So the gist is that, reputation NFTs shouldn't be "soulbound" to a single address because Vitalik said so, rather linked to an identity whether this is a raw on-chain identity or something more sophisticated like what KILT and Encointer and many others are proposing. "Soulbound tokens" fixed to a signle address make them lack lots of functionality despite what the paper "Decentralized Society: Finding Web3's Soul" says. Specially within more interconnected, modularized and expressive technology already present on other blockchain products than what we find on regular EVM accounts.

With that being said, if you have some concrete ideas about reputation or you also experiment with ideas about on-chain reputation, please don't hesitate to send us and email to our on-chain address. We would love to connect.

— We can't use DOT staking in nomination pool,
I think Parity is working on this specific point

— DOT staking a liquid staking protocol
It's up to the protocols to work on a solution using XCM. Bifrost is working on it on its side.

— or DOT locked for crowdloan. So, a lot of community members can't vote for governance (small holders use often liquid staking or nomination pool).
This one is a true failure i think in the design of the current crowdloan system for sure.

I had thought myself to the exact solutions you propose but they all have downsides that Saxemberg mentions.
As democracy is 1 vote = 1 DOT currently, you can't have a better system.
That's alos how it works for decades in the stock market: 1 share = 1 vote.

So it will always favor whales. 

But at the same time, you have a bigger investment in the ecosytem, so it's normal to have a better decision power than a shrimp guy with 100 DOT.

Really i thought a lot about this dilemna and i don't see any solutions BUT increasing the number of participants in governance (delegation, liquid staking and nomination gov rights etc etc..)