CHIP-53 Decreasing Plot Filter Every 2 Years

I believe he just helps connect people but even if he does sell drives thats not a conflict. I think Chia received an investment from Seagate, or at the very least they work together at times. Conflict would imply a competing company or that he’s profiting off insider knowledge.

Well I would say anybody who is promoting Chia, if they have some connection to selling hard drives? For me that’s questionable ethically speaking. In this person’s case they work for Chia. So to me that’s even less ethical. But that’s just my opinion. I could make a business argument saying that expensive 18TB drives are as stupid investment for Chia. But I’m also not trying to sell or encourage people to buy hard drives where I get some sort of kickback. I mean, obviously if you work for a hard drive company this concept is gold. Especially when you can pump out new hard drives to supply the farmers when it has no ethical green value in principal.

Imagine if Ethereum decided to get a sponsorship by Nvidia. Imagine if Chia started putting the Seagate logo on their website and promotional items. You don’t see how that would start becoming a farce?

Dude, just take a breather…and stop trying to find evil under every rock.

Have a good day

3 Likes

What makes you think they didn’t? AMD was in there

My guy really took one point about somebody’s experience and made it his entire argument when it’s just irrelevant to the topic at hand anyways lmao

1 Like

Your contribution is what? Have troubles making some kind of point? Talent is writing 2 or 3 sentences that make no sense. Own it.

I posted this in github but I thought I would duplicate it here for discussion

Here is how I see it → lets say chia corp goes bankrupt. I do not see considerations for this scenario.

Startup tech companies have a very high likelihood (>90%) of going bankrupt. Ones who have less revenue than expenses are more likely to fail.

This CHIP will send us to a plot filter of 32 and stay there forever.

I suggest doing a plot filter change on a as needed basis and not schedule them as its (extremely) likely chia corp will not be around to reset the plot filter and set it to k33. Chia corp will not have to live w/ the problems they create. We will.

2 Likes

I mean we don’t need chia corp to hard fork, bitcoin maintains itself without a central corporation or satoshi.

One great thing about pools is they provide nodes of central organization so if every pool agreed on something we could get word out pretty easily and get a majority to update.

1 Like

You might be surprised.

Chris, wall of text coming, but tldr is “yes but actually no” with this image

Bitcoin also takes 4 years to agree on a taproot upgrade, and less than 2% of the network uses it, then forks itself into bitcoincash for the privacy extension or bitcoin sv so greg can control the satioshi wallets while roleplaying as someone who invented bitcoin and suing people randomly because of his headcannon.

2ndly pools do not, at all, provide consensus. Pools are more like reward pools, but blocks are (suppose to) be signed by individual nodes. I am aware of your flexfarmer where you sign the block on a users behalf and in that case all flexfarmer users could decide what is valid / refuse to sign, but then it would reduce FF rewards versus main client because a legit block would not be signed. Its not possible to sign a block in a diffrent way, only possible to take a block and refuse to sign it based on arbitrary reasoning. It would have to be 67% of all space keys refusing to sign what they deem inappropriate for a while before the main chain lost enough weight to not be valid and cause a hardfork. Chia can do this easily because the main chia client controls >67% of nodes (right now from what I understand there is no public competing node, and the go node FF uses is closed source)

Either way I agree it is possible to hardfork the chain, but not as easy as you would think. Its not something where the top 10 pools get together and say no. It would have to be 67% of the space all changing their clients at once. (may not need to be 67% of nodes as “whale” enterprise farmers control much more space than retail farmers). Right now chia corp can do it easily because they control virtually all the nodes on their software. So right now we are taking an easy hardfork that will become significantly harder to undo as the network increasingly moves to non CNI clients. Bitcoin is diffrent in this regard because the bitcoin core is controled by the bitcoin core developers and they have a lot of social standing in the community to get people to upgrade nodes. This still makes hardforks take 4+ years to get any adoption whatsoever. Chia wont have this when CNI runs out of money.

1 Like

Latest version of FF allows you to use your own node.

Also I suspect chia farmers are much less set it and forget it the way btc are thanks to us being used to constant updates. That being said, in 5-10 years yes it may be much harder to HF.

1 Like

Oh thats cool, yeah I wish they would just do the one decrease and evaluate each time. this is planning too much in the future in my opinion. Who knows if HDDs will even be magnetic, what if they use glass or something or shingled layers where 100TB drives have amazing sequential reads but lower iops than now. This could force users to make stupid big chia files (10TB) to minimize iops. I feel like this should be a one off not a thing that keeps happening w/o input. Each time make a choice as to what is best for the network.

FF cuts IOPS by like 40% or more so we already have a solution :smiley:
Current Chia farmer is unoptimized and burns your drives+power in comparison to FF. FF isn’t just the Chia farmer without the node its like comparing the old chia plotter to madMax.

The below comparison was taken quite a while ago, FF is even better now.
(For each of the two graphs the left side is before FF the right side is after FF)

1 Like

But then you would have to hard-fork every time, that’s also not optimal.
By doing it for 8 years, it reduces the likelihood of more hardforks in the near future.

I also wonder what this Chip create in terms of soft-fork options like shorten or lengthen the 2 year period. But My understanding of hard vs soft fork is not that great

In the Q&A, they built in options to this CHIP to allow adjusting the filter reduction if it becomes too burdensome. The built in options would only require a soft fork.

So I think they are doing it right. They do one hard fork that will for sure prevent plot grinding.

And they have options to scale back by soft forks in the future if the filter reduction becomes to aggressive.

1 Like

How will they know it becomes too aggressive? When our hard drives start to fail?

You have options,

First off, the first filter reduction goes from 512 to 256. The I/O’s will still be ultra low. So really no need to change anything.

But if your’re concern and when the filter reduction starts to get the I/O above your liking, then simply replot to a higher K size.

The bigger K size vastly reduces I/O’ on hard disks.

So you can plan that in 2 years when the filter is reduced to 128, that you will have replotted your hard drives to K33 or K34 to compensate and your I/O’s will stay the same or be lower depending on what size you plot too.

1 Like

Thanks for that insight and explanation. Not a quick decision. Not sure anyone can trust Chia’s predictions on any of this. They cannot predict. I think whether a person is using SAS vs. non enterprise grade SATA should weigh on the decision. However, would Chia just step to the next K size? Or is there a chance they skip the next K size, and move it up by 2? If they could rule that out by doctrine, I would feel a bit better…

1 Like

First you say Chia Network is not capable of reliably predicting the future and should not try. Then you ask CNI to rule out doing something in the future by doctrine.

Which is it? Do you want CNI to make a permanent rule based on the expectations of the future, or openly accept it’s not possible to set expectations around the future?

6 Likes

You control K size. You do not control the global transition to SSD. You know your technology and farmers don’t. We invest in hardware to provide security to the Chia blockchain. That’s no cost to you. No risk to you financially. It’s on our shoulders.

No disrespect meant to Gene, but this public statement about the global transition to all SSD and therefore Chia need no be concerned about running hard drives into the ground? It’s a complete fallacy. I get from a PR perspective you don’t want and frankly can’t tell people that due to security risks, you need to increase the workload on hard drives and thus increase rate of hard drive failure. It’s easier to say it’s a non issue because everyone will be using SSD in the future. You want to die on that hill? Really?

Surely you can understand what a profound change it would be to increase K size to the people providing security for your blockchain. I would call it a courtesy if you can rule out skipping the next K size and moving it up 2 pegs from K32 to K34.

I personally don’t think that’s asking too much. If Chia ruled it out, it’s better for the farmers. Maybe they can make some tough decisions and won’t make a big mistake by choosing K33. If you cannot or will not rule that out entirely? That doesn’t fill me with confidence. You could just say you can’t predict the future. That would say to me that going from K32 to K34 is possible in your mind. It says to me that anyone deciding to do K33 now is making a mistake. But you think leaving this for farmers to ponder this on their own is reasonable and fair?

I stand by one point and it’s a huge oversight if I’m being honest here. When JM talks about hard drives, he is being very misleading. I’m not sure if it’s intentional or not. You know this. There isn’t one hard drive standard. You can not, nor should not say hard drives get XXX amount of writes or reads a year. You can not, nor should not lump hard drives into one and make blanket statements about lifespan. This is obvious. So why do it? Why would an official from Chia do that?

Consumer is NOT enterprise. NAS is not consumer and it’s not enterprise. External is not NAS nor is it consumer. There is a spectrum of drives, brands and uses. Buying used (cheap) adds even more risk to the farmer in terms of drive longevity. Heat being the biggest factor in all of this.

I bring this up (again) because the plot filter is a drive killer. I wish JM would stop downplaying workload on harddrives and using numbers and data like a blanket statement. As if to say all hard drives are rated for this or that. Under ideal, perfect conditions? Sure. ROI is not Chia’s problem. It’s ours. Our hardware needs to last. Consumer / NAS / Enterprise. SAS / SATA. None of these are equal so why present data like it applies to any of these categories?

Lastly, it’s insulting to talk about “this won’t matter in the future” because SSD will be so cheap in the future and i/o won’t matter anyways. Quite an arrogant thing to say when you consider how much $$$ everyone around the globe has spent and continues to spend on mechanical hard drives to provide security to your blockchain. So I’m sure Gene isn’t saying too bad for those running mechanical drives when we (Chia) decide to decrease the plot filter by greater margins in the future to combat unforeseen security risks to the blockchain. For me you are insulting my intelligence. Hundreds of TB of SSD expense/investment for farmers in order to provide blockchain security? Really?

Edit: Last thing to say on this. Sure, replotting is fast now. Ok, fine. It’s definitely a hassle, right? To replot you need to buy more hardware. Fine. But I wonder. How about rewriting/refilling the hard drives? A yearly event moving forward? I know when my hard drives (not Chia related) failed, it wasn’t when I was using them. It was when I decided to move content off in order to back up. That increased activity workload has always, for me, been the point of failure. So do tell. Replotting onto our previously plotted hard drives. Nothingburger workload? I’m sure nobody can say for certain this will be the last time farmers need to replot onto their drives. K size anyone? It’s not wise to assume everyone is running new enterprise drives. Maybe most bought used consumer grade used hard drives. None of what I’m saying here increases lifespan. It only serves to reduce it.

1 Like