For those who disagree, you may just be talking at cross purposes: there is a difference between valuing the assets provided by the float, valuing those assets in combination with the offsetting float liability and restrictions, and valuing those two along with the (for Berkshire, profitable) insurance operation that provides the float.

I think you are correct to say that a lot of people disagreeing about the value of the insurance float are not addressing these aspects separately. Buffett himself has not helped by saying things like "the float is better than equity", but his explanation is that the float is not just cost-free, it has negative cost, since Berkshire's underwriting has tended to be profitable.

I think it would be clearer for him to say that the float is valuable, although it is offset by an equal liability, because it is like an interest-free loan that is indefinitely renewed, and on top of that, the insurance business is valuable because not only does it produce investable float, but it also produces underwriting gains, which can be valued separately. But of course, you are not allowed to count those gains twice, first as underwriting profits, and then as negative cost of float; these two are the same thing.

So when Buffett says that he wouldn't accept $1m for a million dollar's worth of float, what he really means is that he wouldn't accept $1m for the insurance business that produces $1m in float. That is because that insurance business is spinning of $30,000 in underwriting gains every year on average (at a CR of 97), PLUS it can be invested, much of it with good returns. Counting these things separately, say as $450,000 worth of underwriting business (at a 15x pre-tax multiple,) and another $700,000 (using Jim's estimate) based on how the float can be invested, is how you get more value than $1 million, but only part of that is from the float.