I would phrase this as selling 4% of each position you own each year. Whether you choose to measure 4% in share count or current market value at the time of sale, it works out to the same thing, right?

Sure.
The only reason I phrased it as share count is because that number is precisely predictable in advance forever into the future, and emphasizes that the sale quantity is entirely mechanistic. For example, you could put in all the broker orders for the next 20 years in advance.

You can do a forecast of the stock price separately to estimate how much money that raises each period. Prices are hard to predict, but easier than trying to visualize your portfolio value trajectory incorporating a sales schedule at the same time.

It's simply not possible to design a plan that lets one spend the capital while guaranteeing to fund a lifespan of unknown length, while having withdrawals of a predetermined [real] size.
Your choices are having a (probably unacceptable) risk of running out of money, having withdrawals that vary in size somehow. So the main discussion is about what formula to use for the variation in withdrawals.
(You can also have withdrawals that are trivially small and basically have no chance of eating into the capital, but then of course you don't get to spend the capital)

The best solution is to dodge the problem. Cut off the tail of the longevity risk with a pooled scheme like an annuity.

Jim