Actually, optimal taxation models tend to show that a base transfer at zero income with an initially "high" marginal rate for clawing back the transfer at the lowest end (but still no higher than 100%, and falling very quickly as earnings increase!) works very well. (Marginal rates then rise progressively for higher incomes, and paradoxically become lower again at the very top end, trending towards zero at the extreme top-end of the scale. This is actually a consistent result in non-linear incomes taxation models; the very top earner should face zero marginal tax on the very last cent she earns!)
The intuition is that you only "feel" the high marginal rate if you earn very little, which impacts very few people; but the effect of making the break-even point more manageable with lower marginal rates for most earners is felt throughout the incomes scale. What really screws up things is higher-than-100% marginal rates, which are regrettably common in real-world systems amd completely useless.
I'd be curious to read more about that, because none of what you've written makes any sense to me right now. (Why would we want zero marginal tax on top earner who already get zero marginal utility for her money in the first place ?).