Pierre explains PARSEC

I think all that’s needed is the SAFE Network logo in a corner. I like the slides personally.

5 Likes

Slide 32, video 21:00

With this context of interesting gossip events in the gossip graph, the Byzantine consensus can be expressed as follows:

As a node, looking at its gossip graph, I may see a number of interesting gossip events.

I am asked to pick the next interesting gossip event to be agreed on.

If I feel I don’t have enough information yet, I can simply pass my turn.
I will then try that game again after receiving more gossip.
However, if I give an answer, I must be sure that any other node that is not Byzantine (that’s any other good guy, if you prefer) will pick the interesting gossip events in the same order.

Note that once I pick an interesting gossip event and decide it to be the next one, it becomes boring as new shiny things are always more exciting.

I am asked to pick the next interesting gossip event to be agreed on. - What are the criteria for this decision?
EDIT: I may have found the answer to this one, meta votes. So a meta vote is a non-binding test to find the answer to this question “Does my oldest observer strongly see an interesting event that was created by this node?” (Slide 40)

However, if I give an answer, I must be sure that any other node that is not Byzantine (that’s any other good guy, if you prefer) will pick the interesting gossip events in the same order. - How?

Note that once I pick an interesting gossip event and decide it to be the next one, it becomes boring as new shiny things are always more exciting. - Which means that only one node gets to pick the next interesting event I presume?

1 Like

Hi @JPL,
This is a high level description of the concept of a consensus protocol described in terms of gossip.
The next interesting gossip event would be the outcome of the consensus protocol.
“How?” By following all the steps described in the rest of the video: pick observers, analyse their meta votes, hold N parallel meta elections over gossip, apply the concrete coin protocol until you’ve got agreement on the meta votes, then based on the meta votes that were agreed true, decide which interesting value should be the latest to be consensused.
“Which means only one node”… No. The gossip event becomes boring to the node that was able to reach consensus. Other nodes may still find it interesting. In fact, they will until they reach consensus.

3 Likes

Any client or public deliverable should have official branding. I agree that this video should be swimming in MAID branding. Every slide or a background template needs some maidsafe blue. It does not have to be crazy as sometimes clean and simple is best.

Might be best to move this out of this thread…

Anyone want to upvote my comment on this video? It includes a link to the PARSEC video.

I think you’ve been deleted!

Interesting. I just checked. Its still there, top comment.
Shadow banned? haha. It was a reasonable comment.

I can’t see it. Try logging out or opening in a private window.

Yea… it’s not there.

I think they may have “review before post” activated on their videos. Lets see if it gets through over the next few days.

1 Like

It’s because of the link. Default it will be flagged as spam. As poster you can still see it. The good thing is you might get the attention of the owner as they have to approve your post

3 Likes

Easter Egg Hunt

I’ve been tipped off that there’s an Easter Egg :egg: hidden somewhere in @pierrechevalier83’s slides http://bit.ly/parsec-slides .

Students of Latin will be at an advantage, but not too much of one. First to find the five magic words wins a special prize*

A feeling of smug superiority that’s yours to keep all day

3 Likes

Are you talking about the Lorem ipsum dolor sit amet slide?

@dugcampbell @pierrechevalier83

Also, I had a look at the slides from Pierre and I took the afternoon and evening to make it look cool. Not completely happy with the result due to timing constraints (fonts are not maidsafe fonts, probably some small bugs in transitions) but here is what I came up with so far.

Unfortunately I do not have a lot of time in the coming week so maybe/hopefully other powerpoint guru’s in the forum can help and step in. If not, it’ll have to wait a bit. Hopefully it’ll be the first :slight_smile:

//Edit: Updates to the powerpoint

17 Likes

This is awesome @Eureka18 and obviously no offense to @pierrechevalier83 or the marketing team but the slides presented here are definitely more inline with the UX guidelines and overall feel. Props this is very nice. Hope the marketing team pushes these changes when you’re done.

5 Likes

Speaking from experience are we.

Latin is all greek to me so I have no hope

You did a great job on this!

2 Likes

There is around 724 views as of now on youtube. I personally started watching it 4 times and still have not finished it. Its like a great film during which you fell asleep in the cinema. I believe it was supposed to be a simple explanation. So am I just too stupid . Most of you understood it first go?
Thanks

It is a simple explanation of a complex topic(s) if that helps. Not everyone will “get” consensus and certainly not everyone will get the deeper understanding of ABA → full consensus. However, there should be enough for everyone, some high level, some detail. You probably do not need to take the whole lot in and certainly not at once :wink:

8 Likes

RE: poking about PARSEC, I have to say that I couldn’t find any item of concern to me, but I’m certainly no expert. So I forwarded it to those who are: a cryptography professor at one of the top universities in the world, an author of one of the key algorithms on which the safenetwork is partially based, and someone who has offered one of the best dissections of the hashgraph algorithm.
At the very least, PARSEC hasn’t been dismissed out of hand. It says a lot about the protocol that people are taking time to dissect it. Will circle back if/when there’s more.

26 Likes

phew - - I’m getting there slowly, very slowly, and I had the advantage of being reasonably familiar with Silvio Micalis work on Algorand.

3 Likes

Ha ha - no I was tipped off. And I didn’t go to the sort of school where they teach Latin. A knowledge of decentralisation projects is enough to find the easter egg though.

Might be…

1 Like