BuildrSpace

#19. Chia, the best damn blockchain.

• Tarun, Matt • Season 2 • Episode 19

Share episode

Jonmichael Hands,  VP Storage at Chia talks about how Chia's energy efficient blockchain raises the stake for nakamoto consensus and proof of work . 
Before joining  Chia, Jonmichael spent the last ten years at Intel in the Non-Volatile Memory Solutions group. In addition, he served as the chair for NVM Express (NVMe), SNIA (Storage Networking Industry Association) SSD special interest group and  Open Compute Project for open storage hardware innovation.
twitter handle :  https://twitter.com/lebanonjon
Chia Keybase  :  https://keybase.io/team/chia_network.public  as @Storage_jm 
Read Chia how to guides  on :  https://chiadecentral.com/

Support the show

That's kind of the magic of chia is that hard drives are really power efficient be a really exciting part about chia is you don't need some like fancy high end equipment to, you know, be partaking in the network. All you need is a hard drive and a desktop and you just go on your merry way. When the media was very sensationalizing about early on in their life oh my god, you know, chia will wear out your consumer SSD in like 10 days. funny Bram and I wrote extensively on this topic so infuriating because there are these things called Data Center SSDs is you can buy them used on eBay for $100 bucks or $200 bucks that have 10s of petabytes of, of write endurance where you can write to before it wears out sidebar proof of steak proof of steak sucks, true believers and proof of steak won't tell you that. But that's why these proof of steak systems don't have 10s of 1000s or hundreds of 1000s of validators. John, I would say you're wrong. Vitalik is our God. Hahaha. Hey, guys, today I have the opportunity to interview John Michaels. He's the VP of storage at Chia. Before I take you to the interview, I just want to clarify that this podcast series is a passion project of mine and my buddy Matt, which we started out of curiosity. I believe we've been very fortunate to meet all our crypto heroes talk with them been an amazing journey so far. Also, thank you everyone for their support. It keeps me going. Alright, without any further adieu, let's now hear from John. I spent the last 10 years at Intel as a product manager for the data center, NVMe SSDs. At Intel, I joined chin in may, which was pretty wild. But I actually was participating in the chia right right around the end of alpha, kind of like the middle of 2020 like June June ish, which is kind of funny, I thought I thought more people from the storage industry would be interested in you know, a crypto based on storage. You know, what I saw was founded by Bram Cohen. I was obviously a longtime a BitTorrent user, and then the college dorm room. So I was certainly drawn into it. And so, you know, I kind of helped out and join the community, just as a big farmer, you know, I was helping out on writing the wikis for plotting and farming and helping people out with SSDs. In the beginning, the plotters, you know, it's funny now that the plotting is, you know, hundreds of times if, you know, basically more efficient than it was, you know, in the alpha, and, you know, 10s of times more efficient than it was kind of at the beginning of the main net launch. So, but yeah, that was more a lot of the what my background and expertise in SSDs kind of helped out the team early when they were developing the plotters. Great, so John, by the way. So what is, first of all, what does mining look like in chia? And what is plotting that you just mentioned? And I've also, for the sake of our audience, I'm saying there's plotting and farming. So if you can sort of, decipher the two of these terms that are specific to to project Yeah, mining is a bad word it chia. We don't we don't like the word mining. And it seems kind of ridiculous at first saying like, Okay, why did we invent all this new terminology for farming and plotting, but really is it's a brand new consensus. So I think most of you guys may know, within a chia that we use something called Proof of space and time, which is basically using an allocated harddrive space and you're using that to secure the network. And you're using this as opposed to using like CPU or hashing resources, like you do in proof of work. And there's something kind of magical about this, this farming, which is so so one, the whole idea of like, why this needed to be invented was we needed a more energy conscious model for cryptocurrencies that didn't sacrifice on security. So, you know, there's a lot of talk about proof of stake. And I'm sure we'll go there at some point today, because it's hard. It's hard to talk about consensus and crypto without talking about, you know, the other competing consensus mechanisms, but a lot of proof of stake is touted as kind of this energy efficient alternative to proof of work, but, you know, they've been talking about it, especially etherium, you know, since 2014. And yet to have actually shipped it there are there are many, obviously, you know, you know, chains that are shipping proof of stake, but it doesn't come with all the trade offs and people those are kind of overlooked. So in chia, you know, go back to, you know, again, we don't we don't call it mining, we call it you know, farming, and the other word we use and chia is plotting, which is kind of the act of creating these plots, you're basically initializing the proofs of space into this compact form. Now all a plot is is a file that sits on the hard drive. And once you create these files, you basically just sequentially copy them over to harddrive where they basically just sit, you can think of these plot files, kind of like bingo cards, where the the network, in this case, there's this perfect time, which is the VDF or verifiable delay function that basically call out a challenge. And a farmer basically checks the plot on their hard drive to see if one is supposed to challenge. And in every interval, which we call signage point, if you basically have a proof of space that is better than than the one that was challenged, then you can respond to the network, you can make the next block, and you get your Chia as reward. So this incentivizes people to basically, you know, participate in putting their hard drive space up to the network. And so there's just one time thing we have to do plotting, which is basically create the files. Once they're done, they basically just sit on the hard drive, mostly idle. And that's kind of the magic of chia is that hard drives are really power efficient. And obviously, SSDs are even more power efficient, but they're a little bit more expensive, per per gigabyte. But once your plots are on the hard drive, the hard drive basically just sits there idle. And then the person who has the hard drive space can basically earn rewards for farming. And they can do this in a pool. And I'll talk a little bit about if we want to talk about, you know why the chia pooling protocol is very sophisticated and good, but that the really exciting part about chia is you don't need some like fancy high end equipment to you know, be partake in the network, all you need is a hard drive and a desktop and you just go on your merry way, there's even people that are doing their farming on like a Raspberry Pi 4 So talk about a little bit about plotting. And so from what I understood, plotting is it occupies some space on your disk. It's similar to like, proof of work what you do in Bitcoin. So these are basically cryptography, cryptographic proofs, I believe that just occupy space on your desk, once you you kind of create a plot right? Or stand correctly. Yeah, it's in for those who want to go into the nitty gritty, you know, we have a what's called a proof of space construction document, that's part of our consensus that kind of goes into how these are, are made. But some of the magic of why Chia works is some of the cryptographic and mathematical breakthroughs. One of those was the verifiable delay function, which were the first to ship in the actual production system. And the other that proof of time but the others is proof of space. So it's Bram invented is this nice, proof of space format, where you have these cryptographic hashes, and you lay them out in these certain tables. And then once you lay out all this random data, you basically take that data and you sort it and then compress it. And this is algorithmic compression, into nice, neat, compact format, so that it can't be manipulated. And this process this plotting process is this this is CPU and you know, storage intensive, as far as like iops and disk drives and stuff like that. But once that's done, they never need to do it again, for that file, the file just gets created this 108 gigabyte file, and then it just sits there, and you can just copy it over to whatever, whatever the cheapest possible storage medium is, which today is hard drives. So the fun part of plotting is this is kind of like the fun part where you know, where I come into my expertise, necessities and background and data center systems is the goal of plotting is to try to make as many plots as you can on a given system. And the fun of it is there's there's now three different community community plotters, besides the plotter that we ship with. And so now there's tons of different ways and system combinations where you can do this plotting where you have different amounts of DRAM and temporary SSD and CPU cores. And there's not really a one size fits all solution to this, this is kind of a fun computational problem, because now you're trying to say, hey, how do we make these plots as fast as possible? In begininng of chia, that was really, really important, because, you know, the net space was extremely small, and the difficulty was low. And it was a kind of an arms race to get all the storage plotted. But now, now that the network is over 40 exabytes, you know, it's not really a huge race to get on the platform. So okay, so the plotting occupies space on a hard drive, and it probably on an average, how much does it take to create a plot and farm so in chia, you have farming, which you basically have your plots that live on the hard drive, right. The plotting is just this act of creating the plots and you don't need, you know, hard drive we can most people are using for plotting, they're using an SSD as a temporary storage because SSDs are much faster than hard drives, of course, but you can you can plot directly to hard drive, it's just much much slower. And people prefer to plot it like data center SSDs that have high endurance. You can also if you have a lot of DRAM plot completely in memory. And this is the fastest way to plot but obviously the most expensive because you know DRAM is much more expensive than SSDs or hard drives. So more plots means more chance of winning. It's almost like a bingo card like more cards you have like you increase your chances of winning probability of winning, it's not a sure shot in in blockchain world like it shouldn't be. That's how it's designed. So it increases your probability. So more plots you have, and more space. Maybe it's good thing to step back on, in chia, we have this thing called the net space. And this is our representation of all the storage that's out on the network. And this is very similar to hash rate in like Bitcoin or etherium, or any proof of work chains. The net space is the total amount of capacity out there. Now, your probability of winning is directly proportional to your capacity that you're farming, which have all the sum of all the plots that you have, basically, divided by the total netspace. And so it's very similar to hash rate in that regards where the more storage you have, the higher chance you have of winning. And this is where the economics get really very interesting and different compared to like proof of work where in proof of work, you're basically trading a massive amount of electricity. For the resource, obviously, like miners and graphics cards are expensive, but they also consume a tremendous amount of power, hard drives, really the majority of the cost is harddrive, once you have the data copied harddrive, it sits there idle about five watts, and you know, this is pretty significant difference between like, you know, say like an ant miner that's at three kilowatts, or 3300 Watts, you know, versus a five watt harddrive, it is totally different. So, that's, that's really, the exciting part about chia is that the economics of how the incentives work, now that you have the harddrive space where to keep the storage online, it actually requires a very low amount of power, you know, compared to proof of work. Alright, so that's great. That's great. John, a Chia has been compared to Bitcoin, of course, obviously, because of the consensus mechanism it borrows from Bitcoin, and also, you know, other things that has done on top of it, which are, you know, I would say improvements on it on the proof of work, and consensus or Nakamoto. consensus. So you mentioned energy usage. And, you know, sometimes I hear these proof of stake, folks talk about oh, well, proof of stake solves that. Over proof of work, but does it really? And so that's my first question. And the second question is, you know, then there is a talk about, hey, proof of stake uses, you know, no energy or some something like that. But, you know, if you think about it, everything uses energy, right? Nothing is for free. And this, you know, laws of physics, thermodynamics, whatever you want to call it. So yeah, so those two questions I have, like, I'll stop there. Yeah, it's pretty funny. You know, Chris Dixon, Twitter posted something on Twitter, like the other day, about, you know, reposting something that Solana had posted, they posted something on their website about the cost per transaction of Solana. So this is kind of a silly metric, right, you just basically take the annual energy consumption of the network and you divided by the total amount of transactions that your your network could support. And then you have a theoretical lower bounds on like what the transaction, energy per transaction is, and I'll say, will lead up to kind of why that's kind of a little bit misleading, where proof of stake it definitely is lower energy consumption than everything out there. It's a proof of work, you know, the energy thing is not being blown out of the water, it is absolute disaster. You know, Bitcoin, you know, the Cambridge Bitcoin electricity study. And then I used the economist as kind of the sources, but you know, it goes anywhere from like 130 to 180 terawatt hours per year of electricity use. The all the data centers in the world, you know, are a little over 200 terawatt hours. So just to compare Bitcoin versus that it's like kind of equivalent. That doesn't sound very good, right? So obviously, proof of work has energy problems. And so proof of stake it does solve the energy problem, right. But if you only have 1000, or 2000, validators electricity consumption, the network is absurdly easy to calculate you to say you calculate the number of validators. And in this case, when I say validator, I mean, an actual physical system now in ethereum two eth-2 a validator is like every 32 eth, that you're validating to the network, or staking to the network, but like, you can have multiple those on the same system. And that's kind of disingenuous, I think, right? A validator, I think, is an independent entity who has running a system. And no, so to calculate the energy use of proof of stake, it's really easy. You just add up the number of validators that have physical systems, you kind of profile what that system looks like. And then you look at the energy utilization. And, you know, sum it up, and then look at what that consumes over the course of a year for the annual energy consumption. So that's very low, right? If you only have 1000 validators at that's not very much energy. It's just 1000 servers. It's not it's not a tremendous amount of energy, sorry to interrupt. So when you compare that with say, 400,000 plus validators, which chia currently has, if I'm not mistaken or nodes, spread across and compare that apples To apples or any other project. And so, you know, it'll be interesting to see what those numbers are. Yeah. And so you can't compare. You can't compare power utilization without comparing security. And right if you if you think about like, okay, the market. Now you know that Ethereum and it's number two and Bitcoin is number one cryptocurrency. So the market believes that proof of work is the most secure protocol, just as defined by the market cap of the top two coins, right? The market believes that proof of work is the most secure. And that's where all the money is going today. And you can kind of let the market vote for how these things work. Now, when you're comparing, you know, say, you know, I'll get to we've done, obviously, a tremendous amount of work on, you know, the chia energy consumption, that's part of my day job. But I'll get to there in a sec, when you look at like, in a proof of stake network, you know, you're having a massive trade off for this system that consumes low energy, but you're saying instead of physical systems that people are going to basically buy to secure the network and put real energy or work in proof of work to basically secure the network, you're saying, well, they're gonna put their money at stake. And obviously, that has some massive trade offs, where it's much more challenging to reach consensus, and they have lots of ways to solve this problem. And then each one of them, or each one of the solutions, basically comes comes with its own set of problems. I love basically, Bram did a talk at Stanford a couple years ago on chia, and he had a kind of an introduction. And one of the slides that he has is called the sidebar proof of stake, proof of stake sucks. And then he has three slides of basically go on to talk about, like, how terrible proof of stake is, and all the trade offs that people are ignoring, as far as security and slashing and bonding, and the nothing at stake problem and all these nuances of like, like, you know, in in proof of stake, you need this thing called weak. So weak subjectivity where you basically need to trust some validators in where the Chia its Nakamoto consensus, you basically just look at any peer and decide who has the longest chain, you trust the longest chain, there's no, you know, if you were offline, for a while you come back, you don't need like a trusted set of validators appears, you just compute peer to peer network, get a bunch of peers, and then figure out which was the longest chain and you trust that one. And that's the beauty of Nakamoto consensus. And so when you're looking at, again, proof of stake, of course, it has low energy consumption, because you're trading off the security as far and they won't tell you people that are true believers. And proof of stake won't tell you that. But that's why these proof of stake systems don't have 10s of 1000s, or hundreds of 1000s of validators, they have, you know, single digit hundreds, or, you know, maybe 1000 Couple 1000 validators, it's almost impossible. Because of all these problems, you have to have a professional validator that has the right security model that understands how to do the staking, because if you don't, the coins get slashed. And slashing means they take away all your stake, that's not good. So you really have to be professional to do this stuff. Whereas in chia, you know, if you put up your hard drive to the network, it's no big deal. You know, you anybody can basically do it. John, I would say you're wrong. Vitalik is our God. And very, very sad. I did, I did have to read some very old Vitalik blog posts on my research and proof of stake, you know, which dated back to like, 2014, which is funny, because like, they've been talking about this stuff forever. Our joke is like, you know, when is eth 2 gonna finally ship? It's like, you know, it's one of those things. It's coming. It's coming. Next year, every, every year. Alright, so that's fantastic. You know, so Yeah, another point to your security point that you just made is, and a lot of folks in India also when I talk to them, they say, oh, you know, scaling, scaling, and I'm like, dude, scaling is not the issue here. It's the security is the number one issue. And to me, Gene, Gene brought this point home. He had a statement. the other day, I think I was listening to his talks. And he said, If you go to like the top CEOs, or go to banks and all these companies and tell them, please implement this blockchain, but it's gonna have like, it's gonna have the smart contract bugs, every well bugs, every software has bugs, but still, like it's gonna lose money every two months, three months, four months. And what do you what do you think is going to be the CEOs reaction of this big bank? And of course, they want something that's much much, much more secure, or like lightyears secure than what's existing, right?. So that part is often overlooked in the consumer mindset or the market speculators, which like, oh, scalability, let's jump there. Like, what about security? Yeah. And back to that original question you asked me which was like, what is the energy consumption of chia We wrote up our model at Chia power dot org, that's the model I have, it's pretty easy. Basically just look at the net space, the total amount of storage out there in the world that is allocating their space to the network. And then you basically profile what a different set of farmers looks like as far as what kind of hard drives are using what kind of systems are using. And then you make estimations on the kind of bounded range of the energy consumption of those individual hard drives. And then you sum that all up. And you know, right now we're at about point three terawatt hours of electricity consumption on the chia network, which again, is massively higher than proof of stake, but it is 600 times less than Bitcoin. So what we have is what we think of really a good sweet spot where we have, of course, we have real physical assets that are secure in the network in this case, in chia right now, there's, you know, speaking as like the end of November, there's about 40 exabytes of storage on the network and the net space. And that has about a billion dollars worth of storage. So it's real capex secure in the network. And not only that is it's, you know, the security, basically, in Nakamoto consensus, you need 51% to do you know, these attack, well, to get 51% of that storage, now, you need to go buy 20x bytes, which has a, you know, a, you need an entire data center space worth entire data center worth of space, you need, you know, half a billion dollars, and you need about a year lead time to go pull that off. So, even already, you know, just less than a year into this Chia mainnet, you know, we feel like we're in a pretty good space for network security. The other thing, because it's so easy to be a farmer, all you need is a hard drive in a desktop. You know, we have anywhere from like, 300 to 400,000 nodes in the network. And so this is just, you know, again, we're, you know, 400 400 times larger than most of these proof of stake networks, as far as number of independent validators in our network, we call them farmers. Cool. You know, just to put out there, disclaimer, I am a Chia project fan. And so, particularly because of this reason that you mentioned, and it for me, this statement sums up that, you know, under promise, under promise and overdeliver is is what I've seen chia engineers do time, and again, and I am not gonna totally completely put those eth guys under the bus. I mean, they can worship their god, my God is better technology chops and technical folks. So, you know, I leave it at that, you know, I feel like playing a part in the ecosystem, like being a node validator or a person running full node. I feel like, you know, it's like an empowering thing for the for the, for somebody who wants to take part in this new blockchain ecosystem. For those folks, is it easy to get set up, say, you know, on their regular desktops and old computers? to just get started? Start with the farming process or plotting process? Yeah, absolutely. So where do we start? You know, the only really unique requirement about chia plotting is it's, there's a lot of disk rights, which, you know, the media was very sensationalizing about early on in the right Oh, my God, you know, chia will wear out your consumer SSD in like 10 days, and funny, Bram and I wrote extensively on this topic so infuriating because you there are these things called Data Center SSD is you can buy them use on eBay for 100 bucks or 200 bucks, that have 10s of petabytes of write endurance where you can write to before it wears out. So in chia, it's very, what we recommend to people to get to plotting if they go pick up kind of a used datacenter SSD to start or, you know, kind of a high TBW, that's terabytes written is how SSDs are measured in endurance. But you have to start now, if you're only plotting, like, you know, couple hard drives, now, that doesn't matter, just use whatever SSD you have laying around. But if you're going to be serious in planning, like hundreds of terabytes, then now you kind of need some kind of dedicated SSD for plotting. Now, the plotting performance is basically how fast you make the plots is based on the amount of CPU cores, you have the DRAM and then you need this kind of fast temporary SSD storage. And the more those you have, the faster plots you have. So that's really a question of like, like if you're filling up a couple hard drives, it doesn't really matter don't go do not go out and buy dedicated you know, quitter for this just use what you have laying around to there's no options, people do this, like plotting as a service or cloud plotting where you can, if you have fast enough internet, you can pay like a very little amount. Yeah, it was very high in the beginning of the network, because, you know, obviously there was a lot of competition, people could price these very high but you know, I expect the price to basically download a plot over time to get basically very close to like what the actual energy consumption is. And so if you don't have like all these, you know, high end system laying around for plotting, you want to feel the drive. That's it. So selecting an SSD is something we've been kind of spending a lot of time with what I spend a lot of time trying to educate people on is doing with bram bram had this funny Twitter post, which was like, you know, basically like, you wouldn't take a nonstick pan and you know, clean it with a steel wall and then be like, Hey, I don't know what happened to this is weird, it got ruined you basically selecting the wrong tool for the job like, yes, there, there are many tools out there that exists. If you use like a low end, crappy consumer SSD, that, you know, these are made for, like read heavy workloads. Most people, you know, these SSDs again, included with these like four or$500 laptops, you know, these aren't like super high end SSD. This is just like kind of crappy, low end NAND. And you know that they're just meant to be kind of power efficient and low cost, right, these SSDs are 40 bucks, 30 bucks are included in some of these low end laptops like those are not going to be suitable for for Chia plots. But there are plenty of options for if people want to get super serious about chia and learn more there. We have a ton of wikis and how to guides. There's a ton of information on the web about like, where to start. That's fantastic. I also go to Chia decentral. I think I really like that because it breaks down for me. In fact, you know, a lot of folks in my friend circle who were interested in the project, and they're like, oh, yeah, you know, what, where can I get started? And I was kind of sick of like, telling them to be honest, like how to do this, because I have to like babysit them. And then I'm like, here's the link, I this, this place here goes step by step and talks about it. And they know what they're doing. So, so yeah, that that couple of sites like was Chia decentral. And they're a bunch of other sites that do that. I felt like chia decentral did a great job at like, breaking it down step by step for, you know, little challenge folks. And folks like me, I appreciate that. I know for the for the folks listening, you know, before I joined chia, my friend and I started this blog called chia decentral. And I write, I wrote all about plotting and farming. And really early I had these like, basic how to guides on plotting, and you know, that we would just blew up in March, April, like earlier this year, like our website, just like completely blew up. And then obviously, now, you know, I don't have as much time to write for the website, because I'm actually working on the Chia website. And you know, when I post, like, very long, long form blog posts, and I'm usually posting the chia dotnet. But you know, we still have a YouTube for chia decentral, it's up. And we talk a little bit about plotting and farming and just random tech stuff I'm interested in which funny, so it's really a tarun. And I have this you know, mutual friend, I was just at the Open Compute Global Summit, which is just like hardware event, you know, there's like Facebook and Microsoft and Intel, you guys, these guys are there at this hardware event. And I ran into Summit Puri, who's the CEO and co founder of Liquid. And I remember early in the day, somebody on the keybase was like, oh, has anybody and they posted this video, the Linus tech tips, you know, review of the liquid honey badger for, you know, and somebody said, Hey, somebody has anybody tested plotting on this, what I found out that that person was Tarun. And we're actually going to make that happen. I'm going to talk to him about getting my hands on one of these things, we can do some fun benchmarking stuff on on the liquid system for chia. That will be sweet, I would need some of those plots. Wow. Okay. That's amazing. Yeah, I'm glad that you will make that happen. And I've also heard quite a lot about these Mad Max plotters, John, Michael, it's, I actually haven't got into it. But could you talk a little bit more about it? I know, you have like a complete session on YouTube about it a couple of sessions about it. I'll give it a quick you know, so for now, now, we're getting down into the weeds of for people for the chia enthusiast people that want to look into it. So there are, you know, there was the plotter, which is a member of the the chia plotting is creating these files that we shipped the original, you know, chia software with it wasn't very multi threaded or pipeline, right, we've really focused on getting the cryptography right and can making it actually scale and work as far as like producing reliable plots. And this guy named goes by Madmax. I think his real name is Hugo, basically develop this pipeline to chia plotter that basically multi threaded a bit a ton of the phases for the chip plotting, remember the chip plotting you write out, this random data and then you sort it, that's where all the disk writes happen, and then you compress it. And so these happen in different phases, but basically the Madmax plotter you know we're plotting used to take like, you know, 12 to 14 hours for a single plot. The Mad Max plotter now when like a pretty high end system can produce a plot like say 10 to 15 minutes and I got an average desktop can produce a plot between 20 and 40 minutes is kind of the average time for most people's desktops right now. So, GM, here's the thing that I often get also, like people folks are interested in this project. And to them. Decentralization matters a lot like all this, you know, stories, or the narrative that's been told to us, like a decentralization is better, it's more resilient. It's all of that, and a lot of people buy into that. But then they also go into projects, which are kind of have this hidden centralization. And so, you know, I think for me, the cases like, why should I run this? Why should I create plots? Why should I farm, because if I believe in decentralized world, I'm contributing in my own way to this. And with the expectation that, you know, if I have plots, I'm not always going to win. So I think it's for the folks out there that want to enable that, right. It's like, similar to the electric car enthusiasts who first wanted that electric car, it was discontinued by GM. And then, you know, Tesla came along, and they wanted that. So just wanted to put it out there. Like when people have these questions like, Oh, should I just buy right away? Yeah, go ahead, buy it, if you want. But this is an additional thing, which is kind of makes you a part of this whole blockchain internet community. Anyways, so my rant is over. Love it , we make running a full node pretty easy. You know, right now, the database is, let's say, 10s of gigabytes, you know, it's kind of in like 20 to 30 gigabyte range, if it's compressed, you know, so that's not terrible. You just basically kind of look at whatever desktop you have. And basically, to run a full node, it's very light requirements. Like, if you looked into the requirements for like running into ethereum, full node right now, your mind would explode. Like, it's so bad. Like, and it's funny me saying that I used to run an ethereum, full node in 2017. And you could run a full copy of geth. And it wasn't that bad. But these these other chains have just like, that's the one thing Bitcoin got, right, which was like, you know, we make make sure running a full node is easy, because if decentralization matters, then you can't have it like 900 steps and have super high end system requirements for running a full node. If you want lots of validators. If you want lots of participants in the network for security, you need to make it super easy. The other thing is funny this guy, you know, in, yeah, of course, Twitter battles, like I need to stay off Twitter, because I really should not be like me to argue, argue with trolls, but like somebody who's like, well, you know, we call that I think one of these, you know, proof of stake networks for only having 1000 validators, and they're like, Well, isn't 1000 validators, better than 12 mining pools, and I said, how we solve that problem and chia, because if you're in a pool, we have this thing called the pooling protocol, which is all done on chain, where the pool basically, the participants of the pool of the farmers make their own blocks. And the way it works is you that that pool can play has about 20 blocks to basically claim that reward. And then the farmer gets 1/8 of the farming rewards, and then the pool gets the remaining and then it's distributed most the pool, but the farmer is the one actually making the blocks, you don't actually give up any decentralization by pooling, and now you smooth out the rewards for people that have only have a small amount of capacity, reward them the network, now they don't have to have some trade off. And this is the trade off, like the people's arguments about, you know, basically proof of stake being centralized, because, you know, basically promotes these professional validators, and you know, you want to do you don't want to delegate to some somebody that doesn't know what they're doing. And it basically gets it to the point where the exchanges have all the money because in proof of stake, basically, your stake is the amount of money or the amount of coins you have proportional to the network. So it ends up being these, you know, these big, large institutions, and these exchanges end up being the people that have control over the entire network. And then come proof of work. People argue that all these mining, you know, these these entities that produce the mining equipment and have these massive pools create the centralization around pools, and that's definitely correct, right. You know, we saw like when a China crackdown happened, he went 50% Drop in hash rate in Bitcoin. So there's definitely like a large amount of centralization and pools based on people who have access to these to the latest and greatest, you know, mining hardware. And so, yeah, chia has really solved this problem and saying, like, Okay, we want to have our cake and eat it too. We want decentralization, and we want people to be able to do cool things. We want people to be able to earn small amount of rewards, but there's pooling also incentivizes people with a small amount of storage to join the network. And, you know, it's not like either rewards right now are anywhere between like $1 and $3 per terabyte. per month, depending on how much storage you have, which may not seem like a lot to some people, but if you had the storage laying around, it was underutilized resource, if you had just two hard drives sitting in your shelf, or stack hard drives that were just doing nothing, now you can basically monetize those. And that's precisely what a lot of people have done over the last couple months as an netspace has kind of been hovering around the same price has been kind of now what we believe what I believe to be reaching kind of some some equilibrium, but what we'll see what happens in future, also, you know, it's also great, do it yourself kind of endeavor, you know, in my personal experience, you know, I had done engineering long time ago, I had forgotten a lot of things and, you know, then setting up your own just buying stuff, like from from, from eBay, or like other stuff and just playing around with it. I think this is, it's a great way to even understand and and, you know, I know, people had mentioned earlier, this reminded them of like, the early days of Bitcoin when everybody could run a full load. And so I, for me, it was like a great learning experience. And I'm still learning that there are so many things to, you know, just having a J BOD, or like all these, you know, how do you do the storage properly? How do you maintain it, and the actual, like utilization, like electricity costs, and for me has been just stable it it wasn't that massive, even though I have like a very small farm. But yeah, so the DYI part is also where in countries like India is also kind of going away with, you know, everybody's just watching tick tock videos all day long. So that I mean that exactly. And that was part of what was rewarding about you know, we did Chia decentral, in the beginning, I was writing these guides, I'd have all these people reaching out to me, like, oh my god, I'm a Windows only guy and I finally installed Ubuntu server and I SSH in and I felt like Linux God. Like, it's so funny. They're like, you know, these people are just like, it's so rewarding to hear that people you know, into Chia have taken it on as learning experience to basically get better at system administration and storage and understanding your computer systems and building. It's a lot of fun to, like I mentioned, you know, part of the fun is just like, you know, getting these parts off eBay, building systems, benchmarking, comparing versus others, that's kind of a lot of the fun of the community. You know, the early community was really based off the plotting and farming community as as we grow, chia and you know, we didn't talk a lot about today about all the other awesome stuff that chia is doing. I know you have had other other chia folks on the podcast like Bram and Gene, of course. But you know, we are actual participation in developer ecosystem and, you know, amount of apps that are being developed on top of chia is now going crazy. And starting to ramp up, we just released their chia asset tokens a couple of weeks ago. And there's already a bunch of a couple meme coins at launch. And people are making games there's like Mojo dice, which you can like, do this like gambling on chia. Now, there's all these like, things popping up. And this is just what happens. It takes a long time. You know, once the ecosystem starts, it's really the core technological community that are doing the development. And at now, you'll start to see developers and other people start rolling in to really uncover the magic of what makes Chia so special. You know, this the coinset model, we pick, again, very similar to Bitcoins, UTXO model, we believe has like the more decentralization than Bitcoin, more security of Bitcoin, significantly less energy use, but the programmability of ethereum. But besides the programmability of ethereum, with a secure sense in mind, you can audit it and have it be secure and not lose a bunch of money in the smart contracts. And so we think chia is really something special to leave. So, you know, we are, you know, doing a lot of stuff with the storage vendors. So we kind of forgot the cover this is my job is I run the head of storage. So I basically work with all the storage vendors work with the ecosystem on proof of space and time. So people are creating products around plotting and farming, you know, I'm trying to talk to them and let them know what's up and what people are doing with chia. But it's a lot of fun. And, you know, we have a bunch of deep dives in the in the next coming weeks on hard drives. I've learned quite a bit about hard drives over the last 20 years my backgrounds in SSD. So I've obviously been an SSD expert I've been working in that space for over a decade. But I'm actually learning a ton about hard drives that I didn't know over the last year at year just from working with harddrive vendors. So we've actually written up a couple blog posts that we're really posting here soon on. You know how to monitor smart and hard drives and how to tell if hard drives fails and just cool stuff for people that are like now using this using the drives to be helpful for them on their journey. The End User for chia or the target. You know people who will be running like full nodes could be anybody but at this point, I feel like the DYI community is huge. Then I also see design miners have video content creators who are massive storage which is lying around, that needs to be reused, like all that unused storage or reusing the storage or some for this purpose. And all they have to do is click buttons and just download the, like a UI that sits on their on their computer. I think that those are, I feel like are the target folks for chia. And of course the people who are proponents of decentralization, in its earnest, not just the dissent, decentralization sort of, with with centralization underneath, so am I correct? In my assessment? Or like, is it? I'm kind of off there? Yeah, the, you know, right now, I feel like we have a big a big chunk of the early adopters, you know, with, like, 300 - 400,000 farmers on the network, we've got a big chunk of the DIY, and, you know, early adopters and want to participate in this. We'd like to get more and I think, you know, to get to like a million farm, you know, we're asking those questions on the roadmap of like, you know, what will it take to get to a million farmers, what will it take to get to 2 million farmers? Well, the software needs to be a whole lot easier. Plots need to be way cheaper, and we're working on solving all those problems, the software needs to be like much better, you know, the software still, you know, I think the GUI is pretty good for, you know, what, what it was for the kind of initial launch and we've made some improvements. In the last couple of weeks, we released a light wallet that doesn't require syncing the full node to basically be able to send transactions. So there has been a lot of improvements. But you know, we know that we need to get from like a million to 2 million, you kind of need those kind of like mobile friendly apps you need. You need a bunch of consumer stuff, you need to make it really easy. So one of the things I'll be working on is working with like the NAS vendors on making the plugins much easier. You know, for like a Synology, or like a Qnap NAS to basically run a Chia farm in the background it for people that already have these hard drives laying around and those markets for NAS consumer, NAS is huge. So we know that there's a ton of people out there that have these systems that are laying around and storage, a lot of people are buying more storage than they need. We have a bunch of data that suggests that people are only utilizing about 30% of their external storage devices, like their actual hard drives, which basically means they have 70% of these devices free. And that's perfect target for us for CHIA farming. And so yeah, the education needs to be there. And I feel like you know, your site like chia decentral, and couple of others are doing a great job. And by the way, folks, I'm not getting paid by anybody to say that I just really love this project and like this project. And so this is that's why I reached out to JM also and talk and to other folks as well. So it's a great resource to start like a primer to it and like step by step, you know, the digestible content there is great for any newbie and anybody who's in the stage, any stage of like say CHIA plotting and farming. With that GM, I think we are out of time and I know you have a you know, we both have to get back to our daily work. And this was a fantastic intro to plotting and farming by you. I would love to have you more on the show. When you know when you're not super superduper busy with all the other things you're doing. This was fantastic. Thank you been on the show and I would love to have you back. Yeah, thanks again tarun and you know if anybody wants to ask questions about plotting and farming, you know, we're super active on keybase is where all the chia stuff goes down you can find a link on chia dotnet and you know, we're all in there answering questions on plotting and farming. You can find any of us that work at chia I just we're on there all day anyways for work. So we're pretty responsive and transparent about when people want help. We're huge community of plotting and farming people to help and have any other questions about just . Were there it's been a lot of fun discussion about people excited about the project.