Hacker News Reader

> or example, a home NAS system that may currently only be using a small portion of it's capacity. People can then slowly fill that space overtime and potentially make some money from their unused space.

using my spare disks like this does not make sense when it apparently burns them out very quickly. i doubt i would recoup even the cost of a new drive

Plotting, or generating proofs, is what destroys SSDs, afterwards the plots are stored on disk and aren't read/write intensive. Think of it as printing bingo cards (plotting) and then sticking them in a file cabinet (farming). Your printer consumes ink and electricity, but keeping all your printed off bingo cards only costs you the space the file cabinet takes up and maybe the lights you turn on to look for winning cards when they get called.

So it’s just internet bingo? Does the network do anything useful with all the HDD space it takes up like decentralised storage ?

It prevents transactions from being double-spent, like all cryptocurrency. Whether that's useful is up to you.

This is only really true when plotting to consumer grade SSDs, which are limited in the number of writes that they can support. NAS systems usually contain mechanical HDDs and can handle the write load of Chia without issue.

I think this was missed in this whole issue. I'm not arguing for or against Chia, but basically in order to bootstrap a mining operation faster, miners were generating the initial files on a much faster SSD and then transferring to HDD. The initial bootstrap ("plotting") is very write intensive, and thus destroys the SSD. HDD do not have the same sensitivity to write heavy workloads.

So if one simply plotted very slowly on an HDD with the spare space they had, it would not likely impact the expected lifespan, and they could earn a few pennies for their efforts. However the issue with all of these crypto efforts is that systems naturally trend towards centralization for reasons of efficiency. The best systems have some balance between the two. It's simply not worth it for most people to plot 100gb of free space on their laptop for Chia and earn a few cents per month. So it only makes sense for miners to do at scale.

Yeah plotting on a laptop with a 100GB free is not worth it (especially since they typically have consumer ssds in them), but there is a lot of spare storage space on servers/external drives owned by individuals or companies globally speaking. This global storage distribution that already exists and the fact that no specialized hardware (ASICs) are needed is why Chia has been able to become more decentralized than any other cryptocurrency currently.

It will be interesting to see how things centralize or decentralize going forward. When I think of the lottery I'm always amazed by the number of people that play it even with the extraordinarily low odds of winning that exist. I think there will be some equilibrium that's found for the global number of operating nodes that will be a function on the odds of finding a proof and Chia's price.

> but there is a lot of spare storage space on servers/external drives owned by individuals or companies globally speaking.

Companies have spare storage…but it sure as shit ain’t worth it to them. This is the hubris of most crypto projects: they have no real economic grasp of the problems they jump into. Amazon isn’t going to fuck around with Chia for what amounts to literal dust on their balance sheet. Businesses aren’t going to jeopardize their core competency for a couple of bucks. And the average Joe doesn’t actually have external storage space. Spotify, iCloud, Dropbox, etc. It’s all professional miners. Chia is not decentralized.

> When I think of the lottery I'm always amazed by the number of people that play it even with the extraordinarily low odds of winning that exist.

The payoff for the lottery is HUGE. So even though the EV is negative, people still do it. The payoff for Chia is comparatively small. People won’t (and don’t) care.

> Companies have spare storage…but it sure as shit ain’t worth it to them. This is the hubris of most crypto projects: they have no real economic grasp of the problems they jump into. Amazon isn’t going to fuck around with Chia for what amounts to literal dust on their balance sheet. Businesses aren’t going to jeopardize their core competency for a couple of bucks. And the average Joe doesn’t actually have external storage space. Spotify, iCloud, Dropbox, etc. It’s all professional miners. Chia is not decentralized.

Which companies will adopt it and which ones won't is probably hard to determine from the outside. It should be a fairly objective evaluation though internally for a company in terms of what hardware the company currently has and what it will cost to farm Chia and what future expected value for Chia may be. Chia has somewhere between 300K - 400K full nodes operating on the network. That is objectively more decentralized than Bitcoin that has somewhere on the order of 40K - 80K full nodes the last time I checked. So I would say Chia is the most decentralized network I currently know of.

> The payoff for the lottery is HUGE. So even though the EV is negative, people still do it. The payoff for Chia is comparatively small. People won’t (and don’t) care.

That will be a function of the price of Chia. As Chia's price increases more and more people will want to get into that lottery. A lot more people will consider farming some plots of Chia if one XCH is worth $50K compared to $250.

> Chia has somewhere between 300K - 400K full nodes operating on the network. That is objectively more decentralized than Bitcoin that has somewhere on the order of 40K - 80K full nodes the last time I checked. So I would say Chia is the most decentralized network I currently know of.

And Amazon has millions of nodes. That doesn’t make them decentralized.

We have no idea who owns the nodes, or what the concentration of ownership is for the top X%

Chia's nodes are not all running in the same data centers and owned by a single company like Amazon's. To imply that is disingenuous at best. They are operating all over the globe in independently owned data centers and people's private residences. Geo-location by IP will give a fairly good indication of that, but even just a simple survey of farmers from forums, chat, reddit, etc will give you a feel for the heavy presence throughout Asia, Australia, Europe and the US. Also, you can look at the number of unique pool public keys to get a feel for the ownership decentralization. (https://www.chiaexplorer.com/charts/networkDecentralisation) While not perfect, people do not tend to create lots and lots of pool public keys. Therefore, one can be relatively confident that the network has a high degree of decentralization.

>very write intensive, and thus destroys the SSD

I'm not going to argue about how Chia Coin works, but I'm always skeptical of claims about excessive writes destroying SSDs.

From what I understand, this is mostly unfounded fear based on the fact that earlier MLC cells were specced for a minimum of 10,000 write cycles and people extrapolated 5-10 year life expectancies from that. In practice, though, testing showed that the typical cell could be rewritten around 1,000,000 times, and the controllers would just route around the weaker cells that died early.

Anecdotally, even though consumer SSDs have been mainstream for around a decade now, I have not heard, IRL or online, of a single person complain that their SSD has failed (manufacturing defects such as the Sandforce firmware bug aside). At the same time, I've personally witnessed dozens of HDD failures.

I don't fully understand why fear around consumer SSDs is still a thing, but I suspect it's exactly the same "Reddit expert" social dynamic that drives the advice around thermal paste application.

Do you sources for the 1 million writes(erase) on MLC NAND?

I work with enterprise SSDs and that number is well above anything I've seen or heard, even from the manufacturers.

I can't find a source for my original claim, but based on some real world tests[1] it looks like real-world performance is closer to 50-100k writes.

It seems I misremembered the quantitative data, but the overall qualitative point (SSD Endurance is much, much higher than most "prosumers" believe) is accurate.

[1] https://techreport.com/review/27909/the-ssd-endurance-experi...

Most drives these days are TLC and and consumers are more likely on QLC. I havent even seen an MLC drive for a while now.

The claim comes from Chia devs themselves, who recommend not using consumer-grade drives due to their low endurance.

> The one thing in common will be that you need high endurance, due to the fact that it take almost ~1.8TB of writes using the -e flag or 1.6TB without that option to create a single K=32 plot.

> Endurance is how much data can be written to the SSD before it wears out. In Chia this is important because a plotting SSD will generally be at 100% duty cycle and writing all day.

> A mixed use or high endurance data center or enterprise SSD is the best choice for plotting. Used SSDs with plenty of endurance can be found for a good value on eBay, Craigslist, or similar.

> Consumer NVMe SSDs are generally not recommended due to the lower endurance, and they often employ caching algorithms to faster media (SLC, or single level cell) for great bursty performance. They do not perform well under heavy workload sustained IO. There are very high performance consumer NVMe SSDs that will offer great plotting performance, but the lower rated endurance in TBW will result in a faster wearout.

https://github.com/Chia-Network/chia-blockchain/wiki/SSD-End...

Generally agree, but Chia plotting is REALLY write intensive workload.

That's not how hard drives work

Yes, it's probably too late. Or so I've heard.

Chia is like Bingo. Generating the Bingo cards can put a lot of wear on an SSD, but once you have the bingo cards to fill up whatever space you have, you can play for nearly free.

I haven't done the calculation myself, but I've read that it's no longer worthwhile to generate new cards. If so, that means most of the damage has already been done. It's a sunk cost.

But the people who are already in the game can keep playing indefinitely. If prices are too low to even do that, they could back up to tape and go offline for a while.

I agree. I had 20TB extra space on my NAS, so I spent $100 on an entry-level SSD to do plotting. 200 plots later, it's well-used, but not close to dead. I'll probably put my FFIV game files on it to get faster load times.

In the meantime, my operating cost is running the NAS (which was going to be powered up anyway), and some CPU power to handle the farming (not even noticeable on a Ryzen 3900X).

Chia plotting/farming at hobbyist scale shouldn't wear out anything.

But you're not going to make any money either. According to Chia Calculator 100TiB would generate a whopping $6.17 a day. Just like mining any other crypto it's already out of reach for hobbyists.

https://chiacalculator.com/

Hm. Would be a way to subsidize the price of my NAS expansion though. Only 10TB left makes me want an upgrade to 56 TB more, that's not cheap. Allocating 40TB to chia over a year would cover a good 2.5 * 365 = ~1000 a year. That's a nice subsidy to my NAS setup if all it requires is that I burn out a 1TB SSD.

Your calculations are probably correct, but the whole cryptocurrency fad strikes me as completely ridiculous.

It’s like a farm burning some of their crops to generate a bit of hot air for wind power. Like, you can do it and it gives you that small amount of power but you’re literally destroying the resource and probably cause a lot of waste in the system as a whole.

No, it's not.

Chia is just "proof of prior work" and as such has the same arms race problem as bitcoin. People aren't just buying a few TB, farming it and being content. They're building farming rigs and piling on the storage constantly. It just consumes power AND storage.

Again, it was only cost-effective to build farming rigs in the very beginning. It is no longer cost-effective to do so. Even major companies will eventually have a tough time competing with the amount of storage space that is already deployed in the world. IIRC is around ~7 Zettabytes. That is the big difference with Bitcoin. Chia is making use of a resource that already exists in the world and is well distributed. Bitcoin ASICs require manufacturing and building machines for the specific purpose of mining Bitcoin and then distributing those machines.

It's not really worth using your free space.

I currently have 4TB of free space. I could plot them which would result in an estimated $7.57/month of income per month.

But.... that's not a constant $7.57 each month. Instead, it's a 1.5% chance of winning a single payout of 2 XCH each month. I could reasonably expect to go five years, never winning a single XCH.

Who knows what the value of XCH will be then. There has been a pretty solid downward trend in price since May, contrary to most coins, which started trending up again.

It sort of depends on your process right? If you just want to download the client software and point it at your existing drive and start plotting in the background while doing something else, then I tend to think why not? Worse case scenario you continue to get nothing for the space you currently get nothing for. Also, if you want frequent payouts you can join a pool and then you don't have to worry about going 5 years without getting a reward.

Oh, pooling wasn't available when I last researched the topic.

That changes the calculation significantly.

What matters is whether people are churning through hardware, not the scale of the mining operation. Apparently this churn was a problem in the past but it's not anymore. Is this not correct?

> Chia now has somewhere on the order of 300,000 - 400,000 full nodes operating on the network (already making it more decentralized than Bitcoin) and is approaching 36EiB of space.

Are Bitcoin miners as ready to jump ship to the next thing, if it falls out of favor? I don't think so. Chia is not as popular (it is currently #230 on Coinmarketcap).

I also don't think that the power savings of Chia outweigh the space considerations.

> What people who read this article may not realize is that this is the Chia incentive model actually working properly.

I think that most people commenting here understand that this is the way that proof-of-* systems end up functioning at scale. Participants in these systems are incentivized to cut corners, externalize costs, and generally fuck things up for everyone else.

I definitely want to destroy my Home Nas in a matter of months to make a little money. This makes perfect sense. It seems like this is the energy problems of bitcoin but now it's physical hardware that is the inefficiency? I guess Bitcoin is similar with old miners becoming defunct etc and the cost of video cards. But intentionally running down the life of hard drives is a innovative new feature I guess

The video cards were being used for ETH and altcoins. Thankfully ETH next upgrade is to turn off proof of work.

> I definitely want to destroy my Home Nas in a matter of months to make a little money.

That won't happen unless your NAS is a bunch of consumer grade SSDs, which I hope it is not.

Even consumer grade HDDs aren't built for sustained full load. I had few die in months just because of active torrents reseeding which is lighter load than chia.