ADR009: Non-Interactive Default Rules for Reduced Padding

[nashqueue]

This PR is regarding this Question.

rendered

[MSevey]

Deferring to @liamsi and @adlerjohn

[rootulp]

I don't have any blocking feedback, it's all optional to make the document read easier.

I plan on doing another review later this week to inspect the math closer but don't want to block the PR on my re-review.

[nashqueue]

Thinks i need to adjust the calculation:

• Add parity shares as blue nodes to the calculation
• Lower the size of the nodes above the Row roots as it is not a nmt
• Last Row has more blue nodes that it needs
• Update diagrams and result

[rootulp]

No blocking feedback, still LGTM

[evan-forbes]

Nice! Really great analysis 👍 Left a few suggestions and questions. I'd defer to others before merging, but if this ADR is going stale after a month or so, I am fine merging provided we keep it as Proposed status.

I still need to think about the tradeoffs a bit more, but fwiw I'm leaning towards being in favor of this proposal.

If we are only thinking about bandwidth costs of super light clients, and we are only thinking about 134MB blocks (square size 512) for the foreseeable future. Then in the worst possible case we would only increase super light clients bandwidth by about 11 GB per year. (roughly 2,102,400 blocks per year (4 per minute), max block size each block, and worst size of rollup block each block)

However, if we went with this proposal have a more efficient padding, then we would either have a smaller average square size or be able to post more data. If we have a smaller average square size, then we reduce the bandwidth costs of minimized light clients because they gossip a smaller average DAH.

[nashqueue]

However, if we went with this proposal have a more efficient padding, then we would either have a smaller average square size or be able to post more data. If we have a smaller average square size, then we reduce the bandwidth costs of minimized light clients because they gossip a smaller average DAH.

Unfortunately this might be false considering the Jevons_paradox. It would be the same as making Blockspace bigger to decrease congestion, but more block space could in turn create more demand creating the same congestion.

[evan-forbes]

what are the todos left on this @nashqueue before we merge?

[nashqueue]

what are the todos left on this @nashqueue before we merge?

Nothing to add ship it.