Chia surpasses 1 exabyte of hard drive space
chiaexplorer.com
chiaexplorer.com
Edit: My calculation was based on the initial reward schedule of 64 units per block, but when you take into account the fact that it ramps down to 4 units per block shortly thereafter, it's actually a premine of 21 years: https://www.coindesk.com/5-takeaways-from-chia-networks-new-...
Well, Bitcoin itself is a get rich quick scheme for its founders. So are most of its spinoffs.
Regardless of what you think about Bitcoin, this claim is completely ahistorical.
I wouldn't call it a get rich quick scheme for its founder(s) because I doubt they could have known it would take off the way it did.
But, at this point, I don't know that intent matters. Imagine if Bitcoin really did become a dominant currency... one guy would own 5% of it, a deflationary currency! The "fiat billionaires" don't come anywhere close to that kind of concentration of wealth.
It's nice to see the crypto market maturing and ignoring these people so segregated from every other financial market because they never traded anything else. Low float, "pre-mined" assets are in vogue and represent a more compatible funding model for continued development and personal riches than whatever was happening just a few years ago, let alone through the last decade in crypto.
You should ignore the "egalitarian ideology", or completely take advantage of it by launching a Chia-clone without a premine while you simply provide all the storage space first and get the exact same "insta-mine" result while advertising it to people like you that actually care about no premines. A dozen ways to get the same result, its all marketing.
You only get the same result if you completely monopolize mining for the first twenty-one years. If you control 50% of mining, it would take 119 years to achieve the equivalent. My issue is not that there is a premine. It's the scale of it.
Square Inc creates new shares for Jack Dorsey to sell like every week or so. Continual dilution that the market tolerated. Although fairly unique for the equities space, these are the case studies we are growing up with now.
Definition sometimes are lagging behind ever changing reality, same with "premine".
https://www.chia.net/2021/02/10/chia-businesss-whitepaper.ht...
You are free to agree or disagree with the plan, but need to read it before coming to any conclusions.
Satoshi had the decency to mine his coins alongside everyone else.
Other than that - absolutely nothing
Users have prevailed agaist miners and large business in Bitcon forkwors, which failed to force their agenda on the users. This is at least some empirical proof in support that Bitcoin is not a plutocracy.
Miners have large farms that can be bankrupted by users simply deciding to change the hashing algo. No such option in PoS.
In PoS, largest wallets dictate changes. Liquidity begets more liquidity, aka the rich get richer, and such effect has a centralizing effect on PoS. All PoS project to date have failed.
But if I ever grok the building blocks fully enough, I'd like to see if I can devise a system based on "proof of receipt" in which the act of receiving X coins grants you the ability to "settle" a set of blocks totaling a*X worth of transactions (for some constant a).
Like in other blockchains, the node selected to settle a block of transactions gets to keep some or all of the transaction fees for that block. But instead of continually getting selected by virtue of continuing to hold a large amount of coin, you could only continue to be selected by continuing to receive coin (presumably in return for goods and services).
The transaction fees and the constant $a$ would have to bee chosen such that you couldn't turn a profit by sending lots of money to yourself. But I think that would allow for a fair system with a good incentive structure.
What if the goal isn't to profit from block rewards but to attack the network, though? What if you pass stuff around over and over between tons of addresses you own? (Any person can create an unlimited number of addresses, pretty much instantly.)
Even if by the end the total transaction fees result in a heavy loss being incurred, the attacker would only do it if they think they can gain more from the network attack than they'd lose in fees.
I think a secure blockchain system can never make any kind of assumption that 1 address = 1 person/entity or 1 transaction >= 2 people/entities. I think depending on addresses or transactions in any way is probably a non-starter.
One of the key anti-Sybil features of Proof of Work, Proof of Space, and Proof of Stake is that it solely relies on a single scarce, unfakeable, totally fungible resource. The network cares about only one fungible variable: the total amount of computational power, or hard drive space, or funds in the network. That way, it makes no difference if the attacker has 1 address or 1 million addresses.
I agree that any blockchain has to be resistant to Sybil attacks. And that may be an insurmountable obstacle for my idea but, so far, I haven't let that stop me from thinking about it.
There do seem to be some properties of PoW/PoS related to the level of iniquity in control of the blockchain and to their incentive structures, and I think it is worth casting a wide net in search of something better (even if it may not exist).
It seems like everything might be a variation of one of two possible categories: economic game theory (Proof of Stake), or proof of committal of scarce, fungible physical resources. Everything else I can think of just seems like it'd be some form of centralization/oligarchy. But who knows?
(Proof of Stake is sometimes accused of being oligarchical, too, but I think it's still kind of the lesser of all evils, probably.)
Users have prevailed in Bitcoin forkwars over miners and many large exchanges and businesses.
Miners are really more similar to employees of the network, that order transaction in exchange for fees. They are not masters of the protocol, and cannot dictate its rules.
Proof of Work has a nuclear option (change of hashing algo) that would bankrupt the miners and render their farms useless. No such option exists in PoS, just political battles, as we've seen in practice.
In PoS users lost in political battles of EOS (collapsed and rewrote the network rules) and Steem (taken over by bad faith staker). Literally all PoS projects failed in some way until today, due to some centralization pressures it produces.
Some Orwellian newspeak right there. "Reduced to zero". Just call it what it is: blacklists.
Blacklists = permissioned access.
Behave or we blacklist your coins. Welcome to Fiat 2.0
It'll be just wildly more political than PoW chains where you can verify the heaviest chain objectively.
The downside is that "functional" isn't exactly a high bar. For example reducing block rewards to encourage deflation would disproportionately help the wealthy so they may push it even if it isn't ideal for the average person.
Turns out humans aren’t perfectly rational, and handing a huge amount of control to a tiny number of rich people is a fantastic way to discover that while any given rich person might be smart and what not, there’s a non trivial chance that their heir is an idiot and will ruin whatever they’re given. This is entertaining for personal fortunes, but not so funny when they control a currency.
We've seen this throughout history.
That's what PoS will do to small holders. They will be nothing but tiny shareholders with no anti-dilutive protections and zero legal recourse.
I'd much rather deal with a centralized entity where at least I have a some body of law I can rely on.
A system like that is definitely not decentralized.
The code is licensed under a FOSS license, I believe.
From where? Individuals or enterprises? With the direction we're headed (SSDs and cloud storage), it's hard to imagine many people with a few terabytes of hard drive space. On the other hand you can get around cheap hard drives at around $15/TB if you buy external, then shuck them so you can put it in a backpane or something. At that price, it's hard to beat the price of a new drive, especially if you factor in the time it takes for the individual to set it up. Enterprises might have better economies of scale, but I'm skeptical that executives would sign off on running arbitrary code on production servers just to earn some cryptocoins.
It also seems likely that flash storage will become cheaper than HDDs in the next ten years, which will start increasing SSD space on drives due to increased competition. Analysis here: https://wikibon.com/qlc-flash-hamrs-hdd/
Storage workloads aren't uniform. Some need to have localized, reliable, consistently performing persistent storage with room to absorb sudden demand spikes (like customer data on cloud clusters). Others can be stored anywhere and moved with ease, with few requirements on latency, redundancy on a global scale, and read-mostly workloads, which means immediate loss of a single local copy is harmless (like YouTube videos). FAANGs have long since figured this out and how to optimize their storage utilization.
There is zero chance one of them is going to up and decide that Chia is a better use of their "spare" (not really) storage capacity if they have anything resembling competent engineering teams. I know Google does.
Source: Xoogler.
While I don’t think too many businesses will buy this idea of using spare storage for “cryptofarming”, unused resources are pretty common.
The price of a cryptocurrency will tend to roughly follow the cost of generating it. In storage, it’s safe to assume the cost will only go down for the same amount of space. And disk space has been growing faster than transistor density for some time now.
So in the end this is going to be just like every other cryptocurrency: people largely deploying exclusive mining (sorry, "farming") capacity and consuming resources doing so.
At least some data centers destroy their old drives for privacy reasons, so they would need to figure out a way around that if they’re going to reuse them offsite.
AFAIK people already buy up used datacenter drives to use elsewhere, so you're just shifting demand around.
SSDs have limited program/erase lifespan, and the proof-of-space algorithms used by Chia eat into that lifespan pretty fast.
Has anyone tried this? Would be interesting to see the benefits.
Somewhere over that couple days I read that you can farm on an HD, but plotting is faster on an SSD, so I'm making a huge presumption that plotting on a ramdisk would be even faster, without the wear on an SSD.
I'm still not clear on Chia. Sometimes I think I understand it, and other times I'm not so sure...
It’s a matrix of processor cores, threads per core, temp drive bandwidth, memory, and temp drive speed. Oh and cost per TiB of plots per day.
Also plotting happens in several stages, and each has a different set of requirements. So timing the running of plots in parallel and staggering starts also comes into it.
Let's say your mining operation brings in $1000/day in revenue. If it's mining bitcoin you might be paying $500/day in electricity bills, leaving you with $500 as profit to reinvest in the business (ie. buy more hardware). If it's mining chia you might be paying $50/day in electricity. That's $450 in less resources wasted, sounds good right? Well no, because any savvy businesman is going to reinvest that $450 into buying more mining hardware, so you can make even more money. In the end any resource you save from lower electricity consumption just gets poured into manufacturing more hardware.
>And they can be used for other purposes besides just farming Chia
This is a mixed bag, because while it's true that the mining hardware can be repurposed, it also means that during boom times, regular consumers have to pay more for their hardware. See for instance, the GPU shortage.
You can also read more about the power consumption of Chia here: https://chiapower.org/
You can also use any old extra disk space you have on your current computers.
Chia does only one thing: replace $1000 spent on electricity with $1000 spent on SSD manufacturing. It’s unclear which one is worse for the environment, but I think it’s likely that SSD manufacturing is worse, given the (real) mining and toxic chemicals involved.
Multipurpose mining hardware would compromise security of the network. An alternative use for the hardware would effectively be a subsidy on mining Bitcoin.
This would imply that the costs of securing Bitcoin are not being paid by Bitcoin hodlers, and are not fully internalised by Bitcoin.
This is similar to the reason that Proof of Work must not have economic uses external to the network.
If such a subsidy were more valuable than rewards intrinsic to the network, any variance in that subsidy would cause variance in the security of the network, with the likely case being reduced security, the network now being dependent on that external use case.
E.g. if the alternative use is valuable, it might incentivise me to acquire enough miners such that I control majority hashrate of the network subsidised by the alternative more profitable use of the hardware. Any attack on the network is then subsidised by the external use case for the hardware.
Bitcoin is a shitty public ledger, the resource consumption is due to the significant collective interest in it, not what it accomplishes.
There's a bajilion serious hosting providers that have proven their mettle.
And when chips come down to it, crypto never is anonymous.
And there is much to be gained.
Any type of centralized service provider is a SPOF. Some companies have vastly better customer service track records than others, but their internal operations are fundamentally non-transparent, and driven by a particular implementation of the profit motive.
Interacting with a centralized service provider requires you to expose yourself to poorly quantifiable risks - hacking, internal abuse, government interventions, datacenter fires - and in ways not strictly required by the task of data storage - such as a breach leaving your data untouched but stealing your CC number or other credentials.
A better candidate for "the ultimate dream" would be cryptoprotocol-based participatory networks becoming more flexible, more efficient, more resilient, and more accessible than traditional institutions. That's still far away, it seems - and Goliath won't give up without a fight anyway.
The challenge is that it requires you to buy different sets of hardware to earn chia, and the high performance plotting hardware is only needed temporarily (while you’re filling up hard drives). It seems a cottage industry of plotting services is already developing so that those that want to farm don’t have to invest in the hardware to plot.
Farming requires very little CPU and slow disks are not a problem.
> the high performance plotting hardware is only needed temporarily
I'm not sure if you're ever done plotting; if the network space keeps increasing, then you need to increase the number of plots you maintain to keep up -- else you'll gradually farm fewer over time.
I think you probably need more plotting hardware in the initial buildout phase though.
Some are already talking about "plotting as a service", selling disks full of plots once they have finished their own plotting. In the spirit of "selling jeans to gold miners", that might be a viable business model.
Reddit also has some good plotting/farming recommendations among the usual chaff.
Chia blocks are every ~20 seconds. Within those 20 seconds, you need to read 7 + 64 different areas of the disk. Some read addresses depend on the last, so in reality you need to do 7 sequential reads, all within the block time to get any reward.
So basically, if your storage medium takes over ~1 second to read data, you're going to be throwing away some rewards, and if it takes over ~10 seconds per read, you will barely get any rewards at all.
Tape drives are out then, and so is optical media unless it's already in a drive.
ETA: from looking at the docs, it seems that the 7 initial reads do access separate "tables". It should be possible to optimize by storing these on lower-latency storage, and even optimize further via parallel farming of many "plots", of which only a few at any given time will ever need to retrieve the 64 blocks denoting a successful proof. ISTM that tape media has real potential here.
Put those 7 on fast media and you'll very quickly know "I might have won". But without the results from the rest of the 64 reads in a couple of seconds, you won't actually win.
Basically the authors saw how much Ethereum/Solidity contracts are hacked because of unintended stateful side effects and decided to make a purely functional LISP-based language where there are none.
One issue that most people advocating for proof-of-stake gleefully ignore is to work through the initial distribution of the coin- can't stake what you don't have. So you either have a sizable premine, you bootstrap your coin using someone else's ledger or you start out with proof-of-work (or like Chia, proof-of-capacity) and then move on to proof-of-stake once there's a stable ecosystem for it. It does seem like Chia comes with a sizable premine, which is also disappointing but not entirely surprising as practically every new cryptocurrency that has come out as of late has had it to "financially bootstrap" its development and what not. I'd also have some choice words about Bram Cohen, but they'd be rather out of scope here.
there's not really any way to tailor this new shitcoin to specialry hardware: if these fuck gains the all too much religious following, it will shred the storage market. there's no hacks, no special optimizations. there will never ever be relief. it will be reality versus their imaginary but profit making fantasy, pratical use versus endless speculative/zero-sum profiteering forever and ever & fuck them fuck that fuck this so so so much. what a horrible curse.
almost none of these cryptocoin nft whatever activities have any premise or use other than trying to get rich. what a disgraceful use. i'd label it all as esoteric but esoterics often have very powerful underlying use & powers: this is almost all a huge fucking sham where the value is nearly entirely an imaginary number, one too many fucks imagine.
In Chia, fairly high error rates are still very profitable.
Current hard disks might see one uncorrectable error every 1,000 TB read from them. Chia would be fine with one error every 1 megabyte.
That difference might allow custom drive firmwares to put far more bits far closer together, leading to cheaper less reliable storage for farming.
It will always be more cost efficient to assemble tens of thousands of the most optimal hard drives/CPU's/ASICS in a big room, bulk ordering them all cheaply, than it will be to DIY it yourself at home.
Any kind of mining will always end up costing exactly the same as the revenue. And that means you and I will make a loss at home.
You could claim "but I'm using unused hard drive space, so it's free", but really you should be thinking "I bought a hard drive bigger than I needed, but I can partially recover the cost with Chia". And besides, in a few months, the 200GB free on your hard drive won't even be worth the effort of installing the software.
So 1 exabyte is $20M. Well within the scope of VC funding for someone who wants to make Chia appear to be a success.
Assume it takes 7 hours to create a plot on high end consumer hardware.
A big 'plotting house' would have dedicated crypto units to do phase 1, then stream the results over the network to a machine with 400 gigs of ram to do the rearrangement and write out a plot file. It should be possible to do that in seconds per 100GB file.
The plotting I don't see is any kind of bottleneck for a big operation. It's the storage of the resulting files.
My comment was responding to my interpretation of your claim to mean a single entity is likely responsible for the majority of the 1EB, and that could be accomplished for $20M. Based on my own observations over the past few months, I believe the network has been bootstrapped by a wide variety of hobbyists and a few larger scale operations, with far more than $20M invested overall.
It just seems so monumentally wasteful. At least with the liked of Filecoin the disks can be used for actual storage.
Does this Proof of Space approach have any real benefits over PoS (Proof of Stake)?
They also argue it's more green because if you don't want to store the plots, you delete them and the storage is still available as normal storage for other purposes. With PoW, you have specialized hardware that isn't useful for other purposes and you burn a lot of electricity all the time securing the network.
With Chia anybody can get started with their spare hardware.
Right now, it is a gold rush and this is leading to shortage of supply especially in China. This is a short term phenomenon as there eventually comes a point where you cannot compete against people who just have spare hardware and therefore $0 upfront investment.
> This is a short term phenomenon as there eventually comes a point where you cannot compete against people who just have spare hardware and therefore $0 upfront investment.
Very quickly people with just a regular amount of spare storage won't have any incentive to bother, compared to those with racks of PiB of drives and dedicated plotting hardware streaming to it. It's already at the point where 1TB of plots would take a year or so to win a reward, and the difficulty is increasing exponentially, so it would actually be far longer.
Perhaps farming pools will eventually make it feasible to make something with spare space, but likely not enough for it to be worth most people's time.
The more storage you invest, the more numbers you have, it's analogous to bingo in a way.
What prevents this?
If multiple people have good enough plots there can be multiple winners for the same challenge!
You can also win with the same plot file multiple times.
because 1EB is a minuscule amount of the hard drive capacity, compared to the total amount produced per year. According to seagate's blogpost[1], they ship at least 500 EB per year. If you add in capacity from WD/toshiba, it's probably closer to 1000EB if not more. That would make chia's consumption 0.1% of produced hard drive capacity.
[1] https://blog.seagate.com/enterprises/seagate-has-shipped-thr...
2. Started with around 120 PByte. Seven weeks later and it'll be already a decimal order of magnitude bigger.
0.1 percent isn't minuscule.
Very much agree here. I purchased about 144TB of raw storage a few weeks ago for future video and data projects, and it's possible to purchase 1PB of raw capacity (with no redundancy, nor anything to connect to) for as cheap as around $13000 USD at retail pricing.
1EB could be acquired for somewhere between $12-$15M. Not chump change, but considering the number of players pre-mining, absolutely nothing in the scheme of things.
-----
What will be interesting is how this increase in mainnet capacity grows over time, and relative to manufactured capacity. 1.1EB of footprint today is nothing ... but we might start to care if it were to grow at a rate of 1.1EB/day or faster, as manufacturers like Seagate, WD, Toshiba, etc. can't instantly respond to such a sustained demand spike. Channel supply could be exhausted relatively quickly.
HDDs have been hit much harder than SSDs, but I’m guessing you didn’t buy 144 TB of SSDs.
High-speed DDR4 ECC memory (DDR-3200 in particular) can also be tough to come by. A fair number of VARs actually limit quantity (usually to 20 modules per SKU), since it's so tough to get. Again, not due to crypto mining, but because of hyperscalers.
Part of the reason for this is because Chia is really easy to farm — anyone with extra space on their hard drives can do it. There are people farming a petabyte with a Raspberry Pi. This was by design to keep the network decentralized and hard to attack — close to the original intention of "one cpu one vote" from bitcoin.
Right now there's no price for Chia since it's not on exchanges and transactions won't start until May 3. But in the short term it's advantageous to buy enterprise-level SSDs because they're the fastest to "plot" with so it seems that in a few markets, especially in China where interest in farming Chia is surging, a few SSD sellers have seen a spike in demand.
For long-term effects on the storage industry it's hard to say what will happen — there's tons of existing storage out in the world already, the storage industry already sells 1+ zettabytes of storage a year, and unlike single-use bitcoin mining hardware hard drives are useful outside of Chia. The Chia company speculates on the impact on the storage industry on page 23 of their business whitepaper: https://www.chia.net/assets/Chia-Business-Whitepaper-2021-02...
6 YEARS PRE-MINE AND CALLING IT "PRE-FARM" IS AN INSULT
I generally don't have a problem with pre-mines and actively participate in coins with premines.
BUT 6 YEARS!?!?! GET REAL DUDES
What prevents this?
Once a farmer wins a block it's theirs, the same way it works with bitcoin. The reward is tied to a public key, not the original storage. Just like in bitcoin once a miner wins a block they don't need to continuously prove that they have x amount of hashing power in the future.
You can learn more about the details of the blockchain and algorithm here: https://www.chia.net/greenpaper/
Ponzis can be built on top of it, just as the internet can be used to sell penis pills and get rich quick scams. But just as those don’t make the internet a sham, crappy coins and get rich quick schemes don’t make decentralized ledgers a scam.
It's not that impressive and it's not a ponzi scheme.
Any currency which is not backed by real resources is a scam. Including fiat in the last decade - albeit fiat is a scam backed by state actors with armies, making it a bit more trustworthy