Nick Szabo  🔑

Indeed the theory is so bad in this area that we need very large engineering margins for safety.

Agree! Though this can be interpreted multiple ways, eg "we need the block size to be big with large safety margins so blocks don't clog up"

No. Clogging up is not a matter of safety (security): it's mainly just a matter of fees.

When fees rise enough to make physical hologram millibitcoins unspendable, that's de-facto loss of funds. That *is* a security failure.

It's an extremely limited failure based on people using a blockchain in a way it could not sustainably be used (i.e. for micropayments).

And it's not in the slightest a systemic security failure. It's completely localized to txs that used crypto in a very unscalable way.

The other faction would say, it's running a node on consumer hardware that's "using crypto in a very unscalable way"

You could say "ah, but reducing node count is a centralization risk"

But was it really ever in any danger, Nick? I can't help but notice that @rogerkver never found the courage to engage you (that I know of).

Yes. Ignorant pumpers who lobby to impose their hype on blockchain design are a continuing major risk to secure integrity of blockchains.

@leashless no one in this space has articulated any defensible security model yet, political footballs are all there is

Sadly the lack of good models does not in the slightest lower the danger. Indeed it greatly increases the margin engineers need to use.

The "profit" of running a node is the ability to verify transactions.

Miners are not even a very relevant actor here; their only role in fee economics is as a sorting algorithm.

The only behaviors available to *users* are reading and writing. Reading = running a node, writing = sending txs.

From an outside view, probability that 1MB is exactly optimal tradeoff from pure tech standpoint is tiny. Either LukeJR is right or BU is.