Bram Cohen Profile picture
19 Oct, 11 tweets, 2 min read
Now might be a good time to explain that Filecoin uses 'proof of replication' while Chia uses 'proof of space' (thread)
Proof of replication is when the network archives data which users request it to, so there's an archival service which the network performs. Filecoin not only uses proof of replication for service auditing but also for mining.
This is a worse version of what's an already a dubious idea. The technical details of why this is and what tradeoffs can be done to mitigate the issues are very technical and involved. Suffice it to say that Filecoin has failed to find the nonexistant sweet spot.
At the highest level the big problem is that there isn't much of a market for dodgy storage which has no credible guarantees of long-term availability, but Filecoin has many more walls to smash through before splattering against that steel one.
This was already well established many years ago with Mojo Nation, which was the project I worked at before I went on to make BitTorrent based on experiences there. You might notice BitTorrent does no storage whatsoever. There are reasons for that.
Mojo Nation isn't going to happen. Stop trying to make Mojo Nation happen.
In any case what we use in Chia is proof of space. We have no remote storage facility whatsoever. If you want to farm Chia (the equivalent of mining) you can do it off any space you want, with hardly any bandwidth required.
You can farm on spare hard drive space on anything, including your laptop. You can farm using a raspberry pi plugging as many drives of as large a size as you can find which will physically plug into it.
You can reclaim the space you're using for farming at any time simply by deleting the plot files. The only part of it which requires non-negligible resources is the initial plotting (because it has to go over the data a few times, \
obviously you can't plot more quickly than the drive can write) and we recently more than doubled the speed of plotting with more improvements on the way.
So please come check us out at chia.net! We won't require you buy a huge rig which doesn't even work in order to farm! Our testnet up out and running, with mainnet coming soon.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Bram Cohen

Bram Cohen Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @bramcohen

30 Sep
I can confirm that Justin Sun is every bit the scumbag this article paints him to be theverge.com/21459906/bitto…
As a side note, it's highly unlikely that anyone at Tron actually has the p2p live streaming code I wrote running at this point. Everybody who worked on it before left, and the current product plans take no advantage of it.
P2P isn't about saving money on running a centralized service, it's about enabling anyone who wants to to run their own service with no budget. If you're already pissing away millions on flailing development you don't worry about the bandwidth bill, especially before usage
Read 6 tweets
21 Jul
So there are *ahem* some 'defi' projects which claim to get you risk-free returns on your Ethereum deposits. Time for a not-so-hot take (thread)
First of all, for those of you saying 'Isn't that, like, the classic red flag for a Ponzi scheme?' Well yeah that's a reasonable instinct, but I'll try to make a case for these things and point out more specific problems
First of all, a stablecoin has a business model: Some fraction of the funds are never redeemed, so whenever deposits are made you can keep some fraction as profit. If nothing else, there's some breakage in the system (but more on that later)
Read 14 tweets
20 Jul
Good article on how completely blinkered Robin DiAngelo is. I have a point of disagreement though. She isn't well meaning, she's a scam artist (thread) theatlantic.com/ideas/archive/…
In her diagnosis every company is incurably racist and the only fix is to pay her and her ilk big bucks to manage the problem in perpetuity. If that isn't a white person exploiting the problems of racism for their own self-enrichment I don't know what is
Not that Corporate America is purely an innocent victim here. She's brought in so companies can get their gold star claiming they're doing something about race and not have to do anything else theatlantic.com/international/…
Read 6 tweets
16 Jul
Today's Twitter hack had a modus operandi more like a disgruntled employee than a professional who does this on a regular basis
For one thing, the scale of the hack was huge, more Twitter itself getting hacked than the individual people, like the kind of access a sysadmin gets as part of their job
Far another, the nature of the attack was very crude, like someone who had read about Twitter and Bitcoin and happened across that access doing the sort of attack they'd think of from reading the news
Read 5 tweets
9 Jul
And now some thoughts on a possible approach to making a deep learning Chess engine (thread)
What I'm about to say is completely speculative and may completely fail, have already been thought of before, or both, but it's an interesting thought I wan to get out
Going over how engines work there are a few misconceptions and a lot of possible mixing and matching which could be done. PUCT looks an awful lot like alpha-beta, and in fact the two could work together by, for example, having a project to generate a Chess opening book by
Read 15 tweets
7 Jul
The other day I discussed possible modifications to the rules of Chess to make it have less draws. Now a related question: What can be done to make Chess more resistant to computers over humans? (thread)
It must be acknowledged up front that with the current state of technology this is essentially hopeless. With the fall of Chess, Arimaa, and Go, we should assume that current deep learning techniques will easily crush any board game with no hidden information and no randomness
But we can still have the goals of changing the rules such that the game feels more positional and less tactical to human players, and somewhat more well-definedly try to make a game which does a better job of favoring neural network over deep learning engines
Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!