It's nice to see an actual Bitcoin focused lecture! I used to be invited to give a Bitcoin compsci lecture at a local technical university and tried to cram everything into a single 2 hour lecture. I don't think people learned much from that back then 😅.

I think I would have structured it the same way 4 yesrs ago, but I am more inclined now to say that consensus is the actually important part to grasp.

Where is this myth coming from that relaxing OP_RETURN standardness increases the resource burden of node operators?

It is starting to get fuzzier, some full time contributors have employment contracts, most have some kind of grant. Most of us did start out as volunteers and sank countless hours into it without any expectation of renumeration. I also don't think being volunteers or not is all that relevant. There will always be some kind of weird relationship between an OSS project's developers and its users and I don't think there is a way around that.

Yes they can, pruned nodes fully download and validate all historical blocks. They just throw the data away once they validated it.

Glad it made it in finally :)

You prune immediately, even during syncing.

If the mempool is not running a service for the miners to get the most profitable chunk of transactions, what is it about?

This won't change storage requirements. Block size remains where it is. Data in OP_RETURNs is not added to the UTXO set, so it might decrease storage requirements actually.

Not true.

Cash on the internet.

No.

Part 5 seems like the most important chapter, maybe do that earlier? Maybe add something about MAST, Taproot, and Schnorr. Could also end with an overview of available software and its common usage.

Hope to expose this in the kernel API eventually.

Flat file header is passing the test, now to write some migration code, benchmark, and differential fuzz.

Got the coins/block validation logic split PR'ed. Made some progress on a flat file header store. Maybe there is hope for stateless, or even sans-io block validation through the kernel library after all :)

Thank you msvc for consistently making c++ coding more miserable.

Yes

Gigantic!

It is very cool, and kind of unites a bunch of ideas that have been floating around for some time. The key insight he had was that you can take the cut through achieved by the utxo cache and apply it on a network wide level. I'm also very much in favour of using our undo data more, and relaying spent UTXOs together with their respective spending boock.

That is very interesting! Iam suprised you can use this more limited API already. Let me know when you have something to show :)