search

Polkadot-Blockchain-Academy / pba-content

An intensive academic program teaching Blockchain, Substrate, and Polkadot.
https://polkadot-blockchain-academy.github.io/pba-content/
MIT License
53 stars 27 forks source link

PBA3 Berkeley - Polkadot Tracking Issue #605

Open shawntabrizi opened 1 year ago

shawntabrizi commented 1 year ago

cc @BradleyOlson64 @Sophia-Gold @JoshOrndorff

Related and needs feedback integrated:

TODO:

shawntabrizi commented 1 year ago

Some ideas for new lesson plan here:

https://github.com/Polkadot-Blockchain-Academy/pba-content/issues/592#issuecomment-1522553910

Another suggestion for a potential new lesson Plan:

JoshOrndorff commented 1 year ago

I like the outline you right above this. I think of the deep dives that you listed, parachain Validation is most important. If we have too much content, I would prefer to demote erasure coding.

BradleyOlson64 commented 1 year ago

I like your outline Shawn. I'm in the process of attempting to match content to people. There are topics we may want to cover in more or less depth based on access to SMEs. Examples:

BradleyOlson64 commented 1 year ago

Tomorrow I'll be trying to refine our schedule, figuring out how many of these topics we have time to cover and making a first pass at assigning them time slots.

shawntabrizi commented 1 year ago

I will tentatively mark the lesson plan in this thread as the layout for the academy.

As for distribution of work:

Obviously people who are not me need to agree to do this work, but this makes sense to me based on knowledge

naterarmstrong commented 1 year ago

Re: Erasure Coding

It is currently part of the cryptography "exotic primitives" section. I can definitely continue to teach it (no math, of course) and give it some context in the cryptography section, as we feel is appropriate. That way, the erasure coding theory would be covered on the cryptography side, and the polkadot module could really just go over the application (and the economic protocols around it).

shawntabrizi commented 1 year ago

@naterarmstrong i think covering erasure encoding is useful and interesting. I think it is fine to go over it in cryptography rather than Polkadot. In the Polkadot section, we should cover how we use erasure encoding, and the impacts it has on the protocol.

For example:

This might be much less than a full presentation worth of content, and thus can be rolled up into some higher level story.

BradleyOlson64 commented 1 year ago

I feel comfortable taking on the execution cores talk. My focus has been async backing for a while now, and elastic scaling/core allocation is one of the big considerations that went into our async backing implementation.

I don't know of any parachain validation (PVF/POV) SME who will be on site at Berkeley. I could consult with said SMEs on my team to create content, which I'd be willing to present.

Also, last academy I presented the OpenGov lecture. So I'm happy to handle that again if needed.

BradleyOlson64 commented 1 year ago

As for exercises, Joshy and I had a discussion yesterday resulting in a tentative plan of action. I'll summarize it briefly here for visibility and feedback:

For context, Andronik and I delivered three exercises and one demonstration for the Buenos Aires academy. These covered:

Joshy had the idea of combining at least the first three of these exercises. Students would each end up running a relay chain validator and a collator for their custom parachain. They would each reserve a paraId, win an auction, and initialize their parachain.

After thinking on it farther I suggest we run the shared relay chain remotely rather than having the students run nodes. Pierre helped Andronik and I do this last time for the slot auction exercise and it worked great. The resulting network would be much more stable and predictable imo. We could still have the students connect a parachain to their own local relay chain as a side task if we want them to experience running a validator.

KiChjang commented 1 year ago

I don't think I'm a suitable person to talk about the message passing protocols unfortunately. I don't know nearly enough of it to explain the concepts behind it. Would anyone from Rob H's team be able to help instead?

shawntabrizi commented 1 year ago

@KiChjang I thought someone on FRAME had been working on updating all the message queuing? Was it Oliver?

@BradleyOlson64 Regarding deep dives, there might be other topics that you might suggest which I have not included simply because I am not aware of the details below the functionality. Feel free to suggest or alter the topics we cover.

I think @JoshOrndorff might be willing to investigate into, and do a deep dive into Parachain Validation. It might be that we don't have an exclusive deep dive on this topic, but instead the details are included in the Parachain Protocol overview.

@BradleyOlson64 how long was your OpenGov presentation?

BradleyOlson64 commented 1 year ago

@shawntabrizi My OpenGov presentation also covered the Fellowship and was about an hour long.

Here are a couple deep dive ideas:

KiChjang commented 1 year ago

@shawntabrizi If you're talking about just messaging queuing, then yes, it's Oliver, but XCMP is more than just the queuing part -- there's also the part where the bytes of the message is actually sent from one location to another, and I have absolutely no knowledge how that works. I am not sure if Oliver has adequate understanding there as well.

JoshOrndorff commented 1 year ago

@naterarmstrong @shawntabrizi I'm re-watching Andronik's presentation about erasure coding, reed solomon codes, etc. I think this detail should absolutely go in cryptography. When we get to polkadot, (after maybe a 5min refresh on erasure coding) explain that it is used to split up PoVs and make sure they are available. IMO we should not have slides about computations modulus a large prime number in the middle of Polkadot. image

This level of detail is a distraction when we're already trying to explain the relatively complicated steps of getting a parachain block finalized. At risk of drifting too far off topic, @naterarmstrong I think there should be an exercise about erasure coding in module 1. Either in class or part of the assignment. At minimum, recommend a crate and have students split up and reassemble some simple strings. Or if you wanna get fancy, maybe have students implement a simple reed solomon encoding (if that is possible).

So back on topic, my point is that we shouldn't be explaining how erasure coding works in the polkadot module. Students should know how it works from mod 1, and we should explain that it is used for PoV availability.

JoshOrndorff commented 1 year ago

I'd like to get a shared understanding of our vision for hands-on exercises in this module. In the Blockchain module, I strived to do 45 min lecture and 45 min hands-on activity alternating for the entire module. I would love to do that here because I think it really helps keep student engagement high, but I'm also open to other ideas.

As Bradley mentioned, I have an idea that I think can compose several good activities into one cohesive module-long activity that can be adjusted for time availability dynamically. I can sketch in this out in as much detail as anyone wants, but for now I'll give a high-level overview.

This also partially addresses the complaints about long compile time by re-using the same network and artifacts throughout the module as much as possible.

This also reminds me that in Shawn's content outline sketched above, there is no mention of cumulus. I think there should be, right?

naterarmstrong commented 1 year ago

@JoshOrndorff Definitely we should pull that detail out of the polkadot unit. I honestly don't think that level of mathy detail really fits anywhere in the academy curriculum, because it's very likely to be completely unhelpful for most students. I may go over (on a high-level) the names of the different options for encodings / commitments / etc and their properties.

I will definitely make erasure coding a larger emphasis in the cryptography unit though. Most likely it'll end up grouped together with Shamir secret sharing or something similar. I added it to the crypto tracking issue, so any follow up will be there.

shawntabrizi commented 1 year ago

@JoshOrndorff A (short) cumulus lecture sounds good. I added it before XCMP.

Would be good to describe at a high level:

As for the exercise, sounds good. All of the items you describe above seem reasonable, but it would help me to frame the "goal" that students should reach at the end of the exercise. Could be as simple as "Have students feel comfortable navigating the live Polkadot network via Polkadot JS and other common tooling". I think it will be good to anchor to some high level goals.

Sophia-Gold commented 1 year ago

I can do "Introduction to the Parachains Protocol" and possibly one other more general talk -- I was thinking something like "Parachains vs. Other Types of Rollups". I'm assuming these would be something along the lines of at most a one hour lecture, including q&a, and no exercises.

For deep dives, I'll suggest we just base it on whatever people present can speak to and have capacity for rather than feeling like we have to cover certain topics. I don't expect students to retain a whole lot about parachain consensus in this time period so don't think there's a reason to force it. Better to pare things down and do it well.

shawntabrizi commented 1 year ago

@Sophia-Gold, @JoshOrndorff had asked to create the introduction to the parachains protocol presentation. See: https://github.com/Polkadot-Blockchain-Academy/pba-content/discussions/595#discussioncomment-5767218

Perhaps you can work together, but generally would have you coordinate ownership and stuff with Joshy.

On the other hand, would love for you to take ownership of "Parachains vs Other Types of Rollups", which I had envisioned to be the very last presentation, and would also cover other key blockchain scaling decisions:

Blockchain Decisions and Scalability

   - Review and discuss other blockchain scaling solutions in the ecosystem
   - ZK and Optimistic Rollups
   - Homogeneous Shards

I think @rphmeier and @pepyakin are among the main experts we have at Parity who can help paint a comprehensive picture here.

Some other resources for you:

For deep dives, I'll suggest we just base it on whatever people present can speak to and have capacity for rather than feeling like we have to cover certain topics

I pretty strongly disagree here. The deep dives should not be some loose "we talk about whatever we can talk about". It should actually cover and go deep into the topics which are most relevant for the students to have a strong understanding of Polkadot from the perspective of a Parachain Developer.

If we do not have the experts we need to present this material, we should accumulate the information we need beforehand, and create new topic experts who can present for the academy.

Sophia-Gold commented 1 year ago

On the other hand, would love for you to take ownership of "Parachains vs Other Types of Rollups", which I had envisioned to be the very last presentation, and would also cover other key blockchain scaling decisions:

Blockchain Decisions and Scalability

   - Review and discuss other blockchain scaling solutions in the ecosystem
   - ZK and Optimistic Rollups
   - Homogeneous Shards

Okay, that sounds good. I was unclear whether you intended these topics to be part of one presentation or several; I think just one makes sense as this would be one of the most general talks, although possibly supported by content from earlier modules. I would also add restaking to the list, although it's not a scaling solution.

This fits nicely with other work I'm doing around marketing some of the features on our roadmap and I've already spoken with Rob recently about the comparison to rollups and restaking. I'm less up to speed on homogeneous sharding solutions, but spent some time yesterday reminding myself of how Plasma worked and reading about Nightshade. That will be good for me too.

BradleyOlson64 commented 1 year ago

Polkadot Module Schedule: V3 (Changes made since V1: Gave Cumulus lecture to Joshy, switched order of Shawn's Ecosystem/Economy lectures, Split Sophia's lecture in two, added the possibility of Rob's contribution in lectures corresponding to his talks from last academy)

Day 1 (Half Day)

Afternoon Lecture - Introduction to Polkadot (1 hr) @shawntabrizi or @rphmeier if on site Exercise - Explore a shared chain (30 m) @JoshOrndorff Coffee Break (15 m) Lecture - Shared Security and Blockspace (1 hr) @shawntabrizi or perhaps @rphmeier if on site Lecture - Introduction to the Parachains Protocol (1 hr) @JoshOrndorff

Day 2

Morning Lecture - Ecosystem / Economy 1, Staking + treasury + other (1 hr) @shawntabrizi Exercise - Nominating, Validating (45 m) @JoshOrndorff Coffee Break (15 m) Lecture - Deep dive: Execution cores (1hr) @BradleyOlson64

Afternoon Lecture - Ecosystem / Economy 2, Governance + Fellowship + other (1 hr) @shawntabrizi Exercise - Referenda, delegation, and voting on class chain (45 m) @JoshOrndorff Coffee Break (15 m) Lecture - Deep dive: Cumulus (1 hr) @JoshOrndorff Could reuse Andronik’s “Introduction to Cumulus” slides Lecture - Deep dive: Asynchronous Backing (1 hr) @BradleyOlson64

Day 3

Morning Lecture - Deep dive: Perspectives on the Chain (1 hr) @BradleyOlson64 Exercise - Adder parachain + register parachains (45 m) @JoshOrndorff Coffee Break (15 m) Lecture - Cross-Chain Message Passing (1 hr) @shawntabrizi or perhaps @rphmeier if on site

Afternoon Lecture - Polkadot vs Other Security/Scaling Part 1 (45 m) @Sophia-Gold Exercise - Upgrade chains to parachains, register through auctions (45 m) @JoshOrndorff Coffee Break (15 m) Lecture - Polkadot vs Other Security/Scaling Part 2 (45 m) @Sophia-Gold Lectures - Guest content. (1-1.5 hr) Perhaps zombienet. Perhaps Gabriele’s Platform Agnostic Bytecode lecture.

Assumptions:

JoshOrndorff commented 1 year ago

Lecture - Deep dive: Cumulus (1 hr) Speaker undecided. Could reuse Andronik’s “Introduction to Cumulus” slides

:raising_hand: Me please!

shawntabrizi commented 1 year ago

I suggest that we extend " Lecture - Polkadot vs Other Security/Scaling (1 hr) @Sophia-Gold"

And turn it into a full class exercise, discussion, etc...

Perhaps even break down into 2x 45 min sessions, covering separately different topics, with opportunities for questions and discussion between them. I feel this is where a lot of value can be gained by students.

I am fine with breaking down the Ecosystem and Economy, but I would switch the order so I can cover first: DOT Token, DOT Utilities, Staking, Treasury, then cover Governance and Fellowship second.

I would also suggest, if it is possible, and getting students really to participate in exercises, to find a way if possible to put DOT tokens in the hands of the students, and have them actually vote for, delegate, nominate, etc on the real polkadot / kusama network.

EDIT: Also will @rphmeier be around during this time? I feel we should integrate him in the lectures.

BradleyOlson64 commented 1 year ago

@JoshOrndorff Glad you’re up to tackle Cumulus 👍

@shawntabrizi Both suggestions sound good to me. The expansion of @Sophia-Gold’s content sounds like a big value add if she’s up for it.

I’ll adjust the schedule post to reflect changes once we hear from her.

shawntabrizi commented 1 year ago

@BradleyOlson64 @Sophia-Gold @pepyakin

Some coordination needed here: https://github.com/Polkadot-Blockchain-Academy/pba-content/issues/610#issuecomment-1549102668

Looks like this matches the topics for the final day, but how we split it up, who speaks, and slots etc need to be decided.

Sophia-Gold commented 1 year ago

I suggest that we extend " Lecture - Polkadot vs Other Security/Scaling (1 hr) @Sophia-Gold"

And turn it into a full class exercise, discussion, etc...

Perhaps even break down into 2x 45 min sessions, covering separately different topics, with opportunities for questions and discussion between them. I feel this is where a lot of value can be gained by students.

I don’t think an exercise makes sense, definitely think more discussion would be valuable, and am ambivalent about doubling the lecture.

I feel like I could cover everything in half an hour and the more I stretch it out the more opinion I’d likely add wrt the trade offs with Polkadot, which could be good or bad. I haven’t watched past lectures yet, so that would probably give me more of an idea.

The break for discussion in between is definitely a benefit of that format, though.

BradleyOlson64 commented 1 year ago

@shawntabrizi

EDIT: Also will @rphmeier be around during this time? I feel we should integrate him in the lectures.

I think his presence mostly depends on whether all loose ends are tied up for Async Backing by that point. He's discovered more necessary Cumulus + Substrate refactors than expected. Also the audits could turn up new issues. The PBA Faculty page has this to say.

{Rob H (Parity-async)} @rphmeier (UCB req. dates: TBD) Not owning any content formally unless he requests to and is on site. Planning on a public facing social event for him to speak at a minimum if he is on site anytime for the PBA.

pepyakin commented 1 year ago

Oh, it seems that whatever I proposed in #610 would fit into "Polkadot vs Other Security/Scaling" or its extension, and that is already claimed by @Sophia-Gold. Haven't had a chance to meet yet, but from some glimpses here & there, there seems to be no need to worry about delivery & material quality.

Now, this was my only contribution idea so far, making my presence, AFAIU, not really required. Don't take that as an attempt to bail though - always happy to share my knowledge, just a bit uncertain about how that would fit into what I can realistically commit to prepare. LMK if I misunderstood the situation, or if you have any suggestions. [^1]

[^1]: FWIW and for whom it may concern, my element handle is @pepyakin:parity.io. I am on hiatus from work rn, but I will try to check element every now and then. I might answer way faster at @pepyakin in telegram if you are fancy that.

shawntabrizi commented 1 year ago

@Sophia-Gold: I could cover everything in half an hour

A little surprising to me as I thought these topics would actually be quite dense, but looking forward to seeing the drafts.

@pepyakin: Now, this was my only contribution idea so far, making my presence, AFAIU, not really required.

Feels like you would be an extremely valuable TA and writer for the contracts and polkadot sections.

Sophia-Gold commented 1 year ago

Oh, it seems that whatever I proposed in #610 would fit into "Polkadot vs Other Security/Scaling" or its extension, and that is already claimed by @Sophia-Gold. Haven't had a chance to meet yet, but from some glimpses here & there, there seems to be no need to worry about delivery & material quality.

Thanks!

@Sophia-Gold: I could cover everything in half an hour

A little surprising to me as I thought these topics would actually be quite dense, but looking forward to seeing the drafts.

Well, I think without actually doing a dry run through my time estimates are mostly bs. If we need to finalize the schedule sooner than later, I'd err on the longer side. As mentioned, I think there are other benefits of splitting it in two.

To give an example of the level of granularity I'm thinking, I wouldn't cover the intricate details of fraud proofs like @pepyakin suggested since the purpose would be a comparison with Polkadot. It would, however, help to know what related topics would be covered in earlier modules. I'm assuming students would be basically familiar with zero knowledge proofs, for example. But what about things like fraud proofs or data availability sampling or different kinds of PoS consensus mechanisms and the theory behind them?

To give more of an idea, I'd break it down by topic something like:

Okay, after writing that out maybe two lectures is better :)

shawntabrizi commented 1 year ago

I think that looks like an awesome list of topics. 🚀

I am sure timing will come as the material develops.

I would only add that these are all pretty deep topics, and it is fair that we mostly cover them surface level, but we must also give students time to digest the material and have conversations.

Honestly really looking forward to the content.

JoshOrndorff commented 1 year ago

@Sophia-Gold I just came across the concept of a Validium https://ethereum.org/en/developers/docs/scaling/validium/ It might fit nicely into your presentation.

At least in terms of categorizing the various scaling solutions and the tradeoffs they make.

Sophia-Gold commented 1 year ago

@Sophia-Gold I just came across the concept of a Validium https://ethereum.org/en/developers/docs/scaling/validium/ It might fit nicely into your presentation.

At least in terms of categorizing the various scaling solutions and the tradeoffs they make.

I was aware of this in a StarkEx context. Iirc STARKs might have been what started the DA discussion since they were too large to also fit the calldata on-chain (not sure what has changed that they can do this now in StarkEx "rollup mode") so I'll definitely mention it for that reason.

rphmeier commented 1 year ago

I think his presence mostly depends on whether all loose ends are tied up for Async Backing by that point. He's discovered more necessary Cumulus + Substrate refactors than expected. Also the audits could turn up new issues. The PBA Faculty page has this to say.

Thanks Bradley - to clarify I haven't made any formal commitment to the PBA this time around beyond coming by and delivering a couple of lectures, which I hope can be the same as or similar to the ones I gave last time around. I also have some timing conflicts as for when I could come by. After 24/25 July I am free to join. I haven't been actively following the discussions and don't intend to, but if there are specific things we should talk about for content please reach me directly via other channels to alert me.

edit @shawntabrizi I see the checklist you made a few comments back and I should be clear about the level of commitment I've made to this iteration of the academy. I'd like to do more but I've been consistent with Pauline & Dan about what I can realistically take on given the other things I'm juggling. All I have committed to do is create content for a couple of lectures and deliver those lectures.

shawntabrizi commented 1 year ago

@rphmeier can you list those lectures and where they fit into the academy?

For example, I assume you will give the introduction to Polkadot lecture?

Happy to plan around you.

BradleyOlson64 commented 1 year ago

Here's what Rob presented on last time:

  1. Introduction to Polkadot, Parachains, and Blockspace (Slides: syllabus/6-Polkadot/1-Introduction_to_Polkadot_and_Blockspace/1-Introduction_to_Polkadot_Blockspace_Slides.md)
  2. Execution Sharding in Polkadot (Slides: syllabus/6-Polkadot/4-Execution-Sharding-Security/Execution_Sharding_and_Security_Slides.md)
  3. Cross Chain Message Passing (Slides: syllabus/6-Polkadot/5-XCMP-and-Blockspace/1-XCMP_Slides.md)
  4. Blockspace: The Product of Polkadot (Slides: syllabus/6-Polkadot/5-XCMP-and-Blockspace/2-Blockspace_Slides.md)

Three of these roughly match up to Shawn's talks, so I added Rob as a possible alternate speaker for those. Would be particularly great for Rob to give his XCMP lecture again, as that fills a hole in our speaker expertise base.

BradleyOlson64 commented 1 year ago

@shawntabrizi I really liked the visuals you created for your shared security slides. What tools did you use to make them? I'd like to use a similar style.

shawntabrizi commented 1 year ago

@BradleyOlson64 as I mentioned here, just Google Slides and Google Draw.

Any other images just pulled from google search, and potentially edited with Photopea.

If you have any requests for assets you would like help with, or for me to create, I can do that. You just need to give me a specification or something.

BradleyOlson64 commented 1 year ago

Ah, my bad! And much appreciated. I’d like to develop some proficiency making visual aids myself for the future. That said, I may take you up on your offer once I’ve decided which concepts need to be captured in each graphic.

Sophia-Gold commented 1 year ago

@shawntabrizi and @BradleyOlson64 I need to schedule my dry run (remotely, two hours in total) and would ideally like both of you there to give feedback as well as anyone else with knowledge in the topic I'm covering.

This should be either the week of the 12th or 19th. I believe you'll both be in Lisbon at some point during that time, so perhaps best if you can propose some dates and times that work and possible other participants. I'm in EDT.

JoshOrndorff commented 1 year ago

I'd also like to see Sophia's dry run, will be in Lisbon, and am in EDT.

BradleyOlson64 commented 1 year ago

@Sophia-Gold I’ll be home from Lisbon on the 16th. Any date from the 17th forward I’d be happy to attend your dry run.

shawntabrizi commented 1 year ago

I am on a similar schedule to everyone else. I think after the 17th is best, or perhaps just have it happen while we are in Lisbon?

Sophia-Gold commented 1 year ago

I am on a similar schedule to everyone else. I think after the 17th is best, or perhaps just have it happen while we are in Lisbon?

I would decide whether you want to do it when you're in Lisbon and also think of other people you might want to attend. If it's then you'd need to do it later in your day.

If it's the following week I can propose some times and send calendar invites. I'll open it up to anyone on my team who wants to attend and also @burdges who I trust to fact check me.

shawntabrizi commented 1 year ago

@Sophia-Gold i propose you pick a time that is best for you, and those that can attend will, and you should do a recording for those that can't.

Otherwise, we should start to include a formal scheduling app with people to mark when they are available. Everything else is too loose.

shawntabrizi commented 1 year ago

Gav Feedback:

Sophia-Gold commented 1 year ago

Gav Feedback:

  • Polkadot provides: validity & availability, (connectivity too) & finality

    • other protocols provide validity without availability, polkadot provides both
    • cosmos has connectivity, but not the others
    • ethereum has validity, but availability is super expensive, no connectivity
    • with ethereum smart contracts and stuff, you really need availability, not just validity
  • Scalability isnt what polkadot is exactly offering

This is somewhat how I'm comparing things in the second part of my talk, which is focused on rollups. Except I say "ordering" instead of "validity". A rollup bridge alone doesn't really provide validity without a healthy network of off-chain block proposers or watchtowers. Both Celestia and Ethereum with Danksharding provide ordering and cheap availability, but not validity. You also have to keep the data available for much longer due to ORU challenge periods.

shawntabrizi commented 1 year ago

@JoshOrndorff Dry Run Feedback:

Things I would love:

image

shawntabrizi commented 1 year ago

@Sophia-Gold feedback from dry run

feedback here Chat and Polls Chat Polls T Tomek Tomek says:Good morning 12:02 N🌄 Nuke 🌄 Nuke 🌄 says:hey all - one sec getting the team here 12:02 J Jeff Jeff says:"reachable in advance" 12:20 me says:I think i would need further elaboration on how Sassafras leads to avoiding MEV? 12:21 T Tomek Tomek says:Could you explain more about "Move Language: STM with linear types"? 12:23 N🌄 Nuke 🌄 Nuke 🌄 says:👍 12:26 J Jeff Jeff says:It's only one tool of many, but.. If your tx need not be gossiped, and bind the tx to that one collator, then there are fewer parties who can use MEV against you. Now if your parachain code can be designed to give the collator less flexibility, like by not having smart contracts. 12:26 T Tomek Tomek says:👍 Tomek says:Thank you. 12:27 me says:i think you should keep restaking 12:30 J Jeff Jeff says:OmniLedger by Bryand Ford, et al. claimed that if you randomly select 600 nodes for each shared, and assume 80% honest overall, then you'd have 2/3rd honest on each shard. We felt this number should be 1000 after a similar computation. Jeff says:You're screwed if you don't have strong randomness, don't put enough on each shard, etc. We do better because we chose the approval checkers after the adversary commits to their attack. 12:34 N Nate Nate says:+1 on keeping restaking 12:35 J Jeff Jeff says:Adaptive corruption would run kinda "backwards" for us, meaning we could check the last three, or last 100, but we'd need to delay finality for this. We'd be insecure if you carried it forward into the future past the backing phase. 12:36 P pep pep says:Polkadot 2.0 is often was dubbed as Polkadot Squared (although that's not technically correct since since then we cubed it and hypercubed) 12:42 me says:I like Polkadot 2.0 better right now, since it does not attach new brands / catch-phrases 12:42 P pep pep says:agreed 12:42 me says:I think an image would be helpful for Polkadot 2.0 slide, especially if topology is different than nested relay chains, which is an image that has gone around in th epast 12:43 J Jeff Jeff says:the paper we want to write on it should be called Elf Flipping, because the first sharding paper should be called Machine Elves, and not the nodes are being flipped between relay chains. Jeff says:The polkadot white paper was basically an optimistic roll up, not sure if we want to tell people this. 12:52 me says:we should tell them, and then tell them how we pivoted me says:and why 12:52 P pep pep says:IMO yes. Eth dominates the public discussion and we need to explain polka in their language to grab the mindshare IMO pep says:IOW we shud explain things their lang 12:53 J Jeff Jeff says:A few short names I uses for polkadot now: byzantine roll up, cut n choose roll up, 12:54 P pep pep says:👏👏 pep says:👏👏 pep says:👍 12:56 me says:sounds great 12:56 N🌄 Nuke 🌄 Nuke 🌄 says:5 min break 🧉 Nuke 🌄 says:👍👍👍 13:01 me says:present 13:01 P pep pep says:👍 pep says:👍👍 13:01 MK Maciej Kris Żyszkiewicz Maciej Kris Żyszkiewicz says:👍 13:01 me says:i think my brain did not understand "sequencing" vs "ordering". Can i get a tangible example? 13:04 T Tomek Tomek says:Maybe for DOT create L2Bits? Hot or not? 13:05 N Nate Nate says:Is it correct(ish) to describe it as below? ordering of blocks -> ordering ordering of transactions within block -> sequencing That's how I understood it 13:05 me says:that is also how i was thinking, but did not really make sense in the context of the slide. would want to look again me says:was there a formal definition of settlement layer before this slide? me says:👍 13:14 N Nate Nate says:A meta-question about the lecture: Is this before, after, or replacing the blockchain scalability history lecture from the previous PBA? 13:18 T Tomek Tomek says:Could you recommend some Github repository that relates to the presentation in the context of Polkadot? 13:21 N🌄 Nuke 🌄 Nuke 🌄 says:the "T" it transparent, not "trusted" in STARK IIUC 13:24 P pep pep says:I guess it's better to say "bisect execution" rather than "bisect state" 13:28 J Jeff Jeff says:30 days. LOL 13:31 Mattia Mattia says:Great talk so far, but sadly have to leave a bit early. 13:38 N🌄 Nuke 🌄 Nuke 🌄 says:👏👏👏 13:41 BO Bradley Olson Bradley Olson says:👏 13:41 me says:might be more than can be done before the next academy, but pictures speak a thousand words me says:would love some illustrations of the protocols being described 13:51 N🌄 Nuke 🌄 Nuke 🌄 says:👏👏👏👏👏 13:53 N Nate Nate says:@shawn there are some reasonable 2d reed-solomon illustrations here: https://xord.com/research/2d-reed-solomon-encoded-merkle-tree-construction/ 13:53 P pep pep says:👏 13:53 J Jeff Jeff says:there is how much you need to download, but seperately there is the undecodable ratio, which is how much you need to hide to make something unavailable 13:53 P pep pep says:👏👏 13:53 J Jeff Jeff says:At a high level, one of our big advantages is that we use very similar byzantine security assumptions throughout. 13:54 N🌄 Nuke 🌄 Nuke 🌄 says:why we do dry runs 😸 13:55 P pep pep says:the last time I checked the loss factor 50% for danksharding. it's as usual up in the air tho and changes all the time 13:55 me says:👍 13:55 J Jeff Jeff says:A lot of these other protocols are trying to have much smaller security assumptions in one place or another one, but this comes with high costs. Jeff says:in particular these 2d RS things have a worse undecodable ration, which increases their costs in some ways. In polkadot, we have huge fraud proofs, but a better undecodable ratio, so our "sampling" is just each validator voting on its own chunk, and this really cheap. 13:57 pep pep says: https://a16zcrypto.com/posts/article/an-overview-of-danksharding-and-a-proposal-for-improvement-of-das/ I guess that's the a16z, did not read yet pep says:pretty fresh pep says:Jeff, does it matter, if the actual costs per transaction is very far from sub 1$ for Polygon zkEVM? pep says:(which supposed to be extremely unfriendly for proving) 14:10 N🌄 Nuke 🌄 Nuke 🌄 says:In general bringing up things like this, and then pointing to resources to learn more and not speaking to it (potentially incorrectly) is a good bet IMHO 14:11 T Tomek Tomek says:👍 14:12 P pep pep says:Yes, that's only proving costs pep says:Also probably without redundancy
Sophia-Gold commented 1 year ago

I've opened a PR with my slides: https://github.com/Polkadot-Blockchain-Academy/pba-content/pull/657