dankrad

> Should something ever change with the way clients implement the necessary methods on a user defined struct for ssz, the cryptography will not need to be re-analysed, for example....

> This will reduce the number of EC evaluations by 8 since the C1 level that contains 5 elements split into two, is no longer evaluated, and two single fields...

> true, for a single-slot leaf to see another leaf inserted right next to it, it would require a 31-byte collision which is highly unlikely. No it does not, actually:...

> The tree reconstruction algorithm from the witness, doesn't know anything about EOAs, contracts, or similar. Just knows about tree keys and values. Thinking about this, we actually want the...

I am starting to get more worried about this change -- the reason is that it implements a version of the block/slot fork choice, but without the mitigations that would...

> There is a constraint on the slot of a parent block as a mitigation to poor network connections, i.e. `parent_block.slot + 1 == head_block.slot`. If a block was missed...

In addition to my previous reservations on this change, I also have a new one: A fork choice change like this should not be taken light-heartedly. In fact, we explicitly...

> Another thing is that this change does not modify fork choice, as in the part of fork choice that determines what the head should be. It modifies the rules...

The idea of the `VersionedHash` is that it enables upgrading the commitment type (for example, to a post-quantum polynomial commitment) without requiring a smart contract upgrade. The PointEvaluationPrecompile simply takes...

I would argue that the min price should be `10**10` instead of `10**8` Wei/byte. This works out to $16/MB vs 16 cents/MB. The reason for this is that the data...