Is it possible to cheat to get coins through fast plot technology?

If someone can use challenge hash as input and plot a file in a very short period of time (such as using GPU for plot, less than 30s), can he get coins?

1 Like

In theory yes which is why plots are minimum k32 (100GiB) though madmax plotter can do 24 minutes we are still some way off 30 seconds.

Maybe with some huge computer cluster?

If this becomes a possibility then Chia will switch to requiring K33 which is twice the size of plot. I’d certainly think about only plotting K33, but if your going to have to replot to pool anyway just plot K33 when pools go live.

Intesting topic about this raised by Hpool.

No this is a trooling post by someone pretending to be Hpool.

This is a good point. Now plots are coming in much faster, I have some traffic of final plots copying, and got some jamming going on. K-33 will space that out better.

Chia will do a Zoom meeting on the topic in the next few days.
OP is asking about cheating to get coins… you can split this up two ways in regards to fast plotting:

  1. create a plot in real-time that will win the challenge and get you XCH guaranteed.
    You’d have to break the math behind Chia and this requires quantum computing. A super fast plotter won’t be able to do this with tech available for the next decade or so.

  2. create a plot in real-time that will always pass the plot filter
    This is where the plotting under 30 comes in, if you can create and write such a plot to disk under 30 seconds you have created a plot that is equal to 512 “normal” plots… because a normal plot only passes filter 1/512 times.

You’d have to keep creating these 512x plots all the time as they need to conform to the Challenge every 10 seconds, and it’s not going to be economical… you can just do 512 normal plots once and store them… versus running a ton of GPUs to create the 512x plot over and over…

but if this 512x plot becomes viable… then it’s still not a guaranteed win… but a 512 times improvement over the chance of a normal plot to win.

1 Like

Chia should push out the pooling protocol before anything else, rather than digressing on these smaller topics.

They have a lot on their plate indeed.
I’d say doing a Zoom meeting to remove doubt in the community is as important as the pool protocol.

Well, the price of XCH will either trend up after it or it won’t.

What they need is a good public channel to bring statements out. Bram is good at math, not communication and PR.

Ppl are now getting info through various dev/ board members twitter, keybase, github etc.

They should just use their chia network twitter, that can then refer to for longer text.
And bring out solid statements responding to current affairs

1 Like

Expanding what ChiaMax said, having a supercomputer able to create in 30 seconds a plot guaranteed to pass the filter is equivalent to farming just around 18TB. That is because it will only be able to enter 1 of each 3 challenges (being generous) , and those 18TB that are approx. 180 plots will be eligible for all the challenges so they are equivalent to 180x3 plots that are eligible to pass 1 of each 3 challenges.

So in the end you have “theoretically” a system that will be ridiculously expensive, will consume tons of energy and will be equivalent to just plotting and farming a single 18TB disk.

Nowadays to have an expected revenue of 2 Chias per month you will need around 12 of those systems that doesn’t even exist and doesn’t look like they will exist for a long time, running 24x7.

Doesn’t look like a big issue.


Doesn’t look like a big issue so far

I guess the world is full of tech people failing to anticipate that things go forwards exponentially not linearly. However we seem several orders of magnitude away from being able to plot a block winning plot within 30 seconds.

yeah looking back at today 10 years from now, who know what the world will look like then. It’s hard to predict how fast things will go but still feel like many people guess on the slow end of things.

To be more precise, we are far away from doing that in a way that is actually cheaper than just having a lot of plots.

  • You could only ever make a plot that passes the plot filter, not one that is guaranteed to win a block.
  • You would only have to plot phase one, the rest is mostly compression
  • So one “super fast phase 1 plot” is still only equal to 512 regular plots on a hdd and those are still much cheaper than a supercomputer you need for your super fast plot.

*wildcard is probably something quantum computer but I don’t want to think about quantum stuff because it always gives me a headache :sweat_smile:

1 Like

And if / when it becomes possible to do phase 1 in less than 30 seconds affordably, they can open the filter to 256 or 128 cutting the power of the attack by a factor of 2 or 4 (thus double to quadruple the cost).

As previously stated all you do is avoid the filter, not guarantee a win. So you’re emulating cheap storage with expensive computing.

Changing the filter won’t change anything, raising k to 33 or higher should help

Here’s what Bram from Chia says on the subject.

Changing the filter makes the CPU powered approach less effective. What you can do in realtime is create a plot that passes the filter. With the current 1/512 filter, each time you can do this you’re emulating 512 plots. If the filter is changed to 128 (which the network/codebase does support), the same technique emulates only 128 plots, which is much less efficient, while only causing a minimal increase of load on real harddrive farmers.

Within five years the price per TiB of SSDs is expected to become cheaper than that of HDDs (see link below). At this point it will make more sense for new plots to be stored on SSDs. Within a few years of this flippening (say by 2030), the vast majority of Chia plots will be on SSDs.

The main reason for the filter is to relieve the latency caused by spinning HDDs. So within a decade, when this sort of attack might become feasible, all the Chia network has to do is reduce the filter or remove it altogether.

If there’s no plot filter, SSDs will run just fine. And you could only emulate a single plot with this grinding attack, making it significantly cheaper to simply store that plot.

Don’t worry, they’ve thought of this. It’s fine.

1 Like