PBA Buenos Aries Notes and Feedback

[shawntabrizi]

• Video at the beginning introduction ceremony would have really liked some subtitles as the audio in the large hallway was hard to hear.

[shawntabrizi]

Module 1

• Add a slide for "Secure communication" right after the list of cryptographic guarantees

• Move existing slide or add a slide for non-repudiation in that list too

• Might be powerful to mention that the logic of hash functions are completely open source and known to the world, but that does not actually make hashing or the properties of hashing any less secure (for a good hash function)

• In the section where gav was generating public keys and stuff, we assume users understand what hex / ss58 representation of bytes looks like. perhaps we should ensure users are familiar with these basic representations of bytes and what that means.

[nuke-web3]

My WIP stuff from BA 2023:

1.1

• Data integrity slide image is Cambridge specific ID card, remove date & place. - Suggest signature and PBA cert without a year.
• Symmetric Encryption: enigma cipher, add a photo?
• update notes on what lossy vs/ lossless is - good question was hard to answer this time on what that is in the mind map
  • could be wrong actually - for example, a many signatures can be valid for the same data IIRC.

Econ

• should discuss explicitly around 2.3 the idea of marignal utility in the context of the question on 50% chance at 1M vs. 90% chante at 300k
• Econ games - will we do them again? If so we should rewrite the backend as it's not ideal for our use case and a little buggy. It works, but it'snot a great experience (talk to Florian about this, rather larger piece of work)

Mod 3

day 1

• emojis added
• fix spacing/padding
• rounded image boarders
• ...

[JoshOrndorff]

many signatures can be valid for the same data IIRC.

indeed https://en.wikipedia.org/wiki/Probabilistic_encryption aka "randomized encryption". This is an important feature because otherwise encrypting the same message multiple times yields the same ciphertext. Imagine we have a standing daily meeting and I send you the message, "Let's delay our 11:00 meeting until noon today." on occasion. If that yields the same ciphertext each time, the adversary would soon notice.

[shawntabrizi]

update notes on what lossy vs/ lossless is

The way i understood the diagram and those notes, is the following:

• One of the primitives of cryptography is one way functions. That is obviously a broad term, and there can be many kinds of one way functions. Most one way functions that we could invent are probably not that useful, however there are at least two which have been specially crafted to have very useful properties.
• Hashes is one example, which explicitly takes advantage of being lossy. Lossy in order to provide a fixed length output, and to obfuscate the data it is meaning the represent. These are properties that we wish to have for a hash function (among many others which are listed in the hashing section), not some kind of unwanted result.
• Asymmetric crypto also uses one way functions, but it explicitly uses a class of them which are lossless. Again, this is not something we can compromise on, since many of the properties of async functions rely on and are useful because of the fact it is lossless. It is not some random coincidence that async functions are lossless, they would be useless otherwise, for example for encryption / decryption.

So in the world of one way functions, these two have been specifically crafted for the features we wish to gain from them, and those features are derived from the fact that they are lossy or lossless.

[shawntabrizi]

Module 3

• Some of the images use a dark purple text / arrows and it is hard to read imo. Just needs to change the color.

[shawntabrizi]

Module 4

• Light green with white text images are hard to read

[shawntabrizi]

Add a section in FRAME to go over pallet::config, things like pallet::constant, configurable get, conversion trait, etc...

[shawntabrizi]

improve final assignment for frame to clarity the "price api" and it NOT being an RPC

[shawntabrizi]

Student Feedback

• Make Crypto + Game Theory sections shorter

• Make FRAME section longer

• more FRAME examples

• Harder rust exam

  • more working with macros
  • more generics
  • more substrate patterns and basics
  • FRAME tutorials mandatory-ish

• Hardware requirement, perhaps some Mac M1 people can borrow

[shawntabrizi]

For the rust exam improvements:

^^^^ unfortunately still seeing a lot of unnecessary unwraps in some of the projects submitted.

Thankfully, some of them are "safe", but just really non-ergonomic use of Rust.

Here is a small primer:

#[derive(Default)]
struct StorageValue {
    value: Option<u32>,
}
impl StorageValue {
    pub fn get(&self) -> Option<u32> {
        self.value
    }

    pub fn set(mut self, v: Option<u32>) {
        self.value = v;
    }

    pub fn put(mut self, v: u32) {
        self.value = Some(v);
    }
}

enum PalletError {
    NoValueSet,
}

fn my_function() -> Result<(), PalletError> {
    let MyStorage = StorageValue::default();

    // Read some storage, if NONE return an error
    let value = MyStorage.get().ok_or(PalletError::NoValueSet)?;

    // Anti-pattern. Don't do this!
    let maybe_value = MyStorage.get();
    if maybe_value.is_none() {
        return Err(PalletError::NoValueSet)
    }
    let value = maybe_value.unwrap();
    // This is at least a little better, but still an anti-pattern
    let value = maybe_value.expect("cannot panic because of `none` check right above. q.e.d");

    // Do some logic only if a storage value exists, but don't return an error
    // which would halt the whole extrinsic.
    if let Some(value) = MyStorage.get() {
        // do things with value
        // If there is no value, then this code will never execute.
    }

    // This is okay... but probably means you should be using `ValueQuery` configuration.
    // Can't really make a simple example here of `ValueQuery`. Check the presentations.
    // So probably again an anti-pattern.
    let value = MyStorage.get().unwrap_or_default();

    Ok(())
}

[shawntabrizi]

I honestly do not see the game theory parts of the course show up in their final projects. Many people making really bad state machines, or ones which are not really good against all malicious actors.

I think we should be much more concrete about designing games for the blockchain in the game theory section.

[shawntabrizi]

Sergey:

• more xcm examples, less reading docs from the slides
• introductory doc with preparation materials for students
• revisit game theory and economics
  • game theory not really used in the projects, although the section was great by itself
  • economics would be useful more in the context of blockchains, parachains, polkadot.
• start frame as early as possible. Most of the basic (cryptography, economics, etc.) content can be either compressed or omitted.
• Cryptography project felt disconnected from the rest.

[nuke-web3]

Update from student Nate Armstrong there who took great notes:

In case anyone wants to go over things we learned in the class that aren't necessarily present in the slides, I collated most of the questions and answers that I took notes on during PBA into a github gist here. It's mostly complete, about 7k words, and hopefully useful! The sections on FRAME and polkadot are by far the most complete.

Kian remarked it also helps us to see which points were ambiguous, and how to improve.

[nuke-web3]

I think the feedback is incorporated or well on it's way to be for UCB.

https://github.com/Polkadot-Blockchain-Academy/pba-content/issues/437#issuecomment-1412339144 I think we will see if the "assignment 0" helps get some things like this more well understood.