Good Lord I love this job.
29 Likes
I’m a fan 
This would make for smoother sailing
8 Likes
Not really, Parsec was a very good algorithm for total order (although very heavy) However we cannot total order a complete network that is so wrong. DAG-based consensus also suffers from a polynomial increase in effort as participants increase. This and some other things became clear with time, in any case, we did put huge effort to try it.
A problem we had with Parsec is folk worked on Parsec and admittedly did not know routing. Then folk worked on routing and did not know what it was being used for (what vaults were doing). So a big disjoint in the big picture. That caused many headaches.
Then the belief you can have concurrent data and consistency was lost on many even though crdt is in wide use.
So the wrong approach that I for years (since 2015) argued against was the notion of control everything (total order the world) and that was such a big hill to climb. Don’t get me started on timers, in code panics and serialising all events to happen 1 at a time 
all fights, all continually implemented and all caused problems and failures.
So basically Engineering wise the control the world is wrong IMO and work with reality (i.e. live with what you cannot control) is correct IMO.
Data chains which was done in 2015 solved many issues but was incomplete. it had code, patents, papers and needed completing, I had hoped by the team (some were keen to look). The missing link was making it a semilattice, which means no merge of sections. Merging sections in any case means you have lost data, so merge was not required and would break the semilattice requirement of a CRDT. The team instead looked at consensus algorithms and were persuaded that was the way to go, the only way and that’s where there are issues. The real answer is that it is a mix of consistent data, concurrency, fork prevention (total order), fork resolution (i.e. multi party doc editing) and more.
For currency AT2 is lighter and faster than Parsec, so we use that there. In other places, we use what we must and hopefully no more. We can also achieve this using determinsitic fork resolution as well, in fact routing uses this for SectionChain “order” (i.e. resolved forks, but these can advance to fork prevention with small effort).
So bottom line total order as an answer to all data type mutations or network actions has never been a route to go. The answers are simpler, smarter and more secure.
In terms of mass exodus, I think the team were great, most were deep thinkers, all of them were good people and some would have loved to work on the code we have now. I know @AndreasF and @bart had interests there, in fact Andreas suggested splitting the section chain from the data chain and guess what, we have. Bart liked the CRDT notion or at least found it interesting.
However the vision is the vision, changing that won’t really wash with me and I know I am as stubborn as hell about that 
It’s all life and lifes story though, we all make mistakes and we all should learn.
42 Likes
I think that shows the power of a good vision too, though. Without it, the project could have ended up in the weeds. Sometimes we need to prove what doesn’t work, in order to discover what does. Obviously, it’s better to get it right the first time, but sometimes that just isn’t possible.
Thanks for sticking with it, David. Also, thanks to the team - previous and current - for getting us here. You will go down in history!
23 Likes
Think this needs reiterating. Thanks all.
14 Likes
Thank you for all your efforts. This is a bit like groundhog day and the day i keep living is the one before Christmas, tomorrow sounds great guys. I really need to get some decent sleep.
4 Likes
I can’t wait to see it in action 
Great answer. Thank you for sharing. You can make this vision happen.
5 Likes
In hindsight it seems lots of time has been wasted because David was right all along on certain choices. It’s also true that nobody knew for sure, and I suspect everyone has learned a lot by exploring the directions that did not work. I always come away with a sense of knowing better after driving deeply into something, whether it works out our not.
I think having explored all these possibilities gives everyone more confidence in the approach arrived at in the end, although it is also bound to evolve, though hopefully in more of an incremental fashion!
19 Likes
Very Steve-Jobs-esque. Visionary leaders often get a lot of flack for having strong convictions—until the vision successfully pans out. Then visionary leaders are hailed as geniuses. One must have the strength to push against the wind until the wind is at your back.
18 Likes
This and I could not explain in a way that was obvious (to most of the team). It’s the “if you cannot explain it easily then you don’t know it well enough” scenario. If I could have been more clear, perhaps it would have been quicker to get here, I guess we will never know, but I suspect lack of clarity did not help the team.
It’s something I feel you need to do, let folk take choices you would not, however hard it is, as long as they share the vision, that works. Or at least I think so
16 Likes
Really enjoying these explanations about the lack of need for total order and control.
As a concept it goes so deep, and could be extended to so many areas of life.
I really hope that whatever is built on top of the network will take a similar approach, and I retain some level of optimism just because I think the structure that things are based on does eventually become significant.
On that note though it’s my pet theory that any social network based on binary code will probably lead to binary opinions!
3 Likes
stolen for Twitter 
3 Likes
Should be Twitter’s tagline!
3 Likes
Will it be better to hold off till next week? It is getting close to the weekend and I recall you saying extra support is required for testnet releases
1 Like
Don’t put evil ideas in their heads.
7 Likes
NO
Mods - instead of these poncy heart “likes” can we get an option for absolutely hating a post, please?
5 Likes
Really nice to see you looking back retrospectively, it is the prelude to the birth of something truly amazing coming this week. It must be a very emotional event after having built the network for 7+ years.
What a level of persistence!
18 Likes
I just thought you would have a few days more for cheap coins. Like me
3 Likes
I’m skint. Even tried to take the cat to the pawn the other day so I could buy more MAID.
1 Like
Look into my eyes, look into my eyes and pawn the 
3 Likes