iddo, I think there are two things to point out with respect to that wizards' conversation (where we didn't really reach consensus):

1) The DoS danger can be avoided with some trivial heuristics, for example: drop or ignore orphaned blocks with less than factor X of the difficulty of the currently accepted best block, and drop blocks furthest from the current best block when orphan storage exceeds Y megabytes/gigabytes. In reality there are a spectrum of possibilities between nodes having omniscient knowledge about all forks (pure GHOST), and nodes blindly following the most-work chain (bitcoin current). Even a heuristic-limited, DoS-safe version somewhere in the middle of that spectrum would be an improvement over today.

2) This is a local-only change that does not require consensus. It's okay for light nodes to still follow the most-work chain. What this change does is provide miners specifically with a higher chance of ending up on the final chain, by better estimating which fork has the most hash power behind it.

3) What this patch actually accomplishes is a little obscured by the hyperbole of the OP. There are alt coins which had a constant difficulty, and those which had incredibly low block intervals. I'm speaking in the past tense they basically imploded when their network size increased and the propagation time approached the block interval. This patch wouldn't fix that problem (because eventually you run out of bandwidth), but it does let you get a lot closer to the network's block propagation time before witnessing catastrophic consequences. This is important because when we increase the maximum transaction processing rate of bitcoin, it will be by either dropping the block interval or increasing the maximum block size (both of which are hard forks), and how far we can safely do that is bounded by this limit, among others.