[ANN] BTC Guild's Mitigation Plan

[eleuthria]

This mitigation plan is no longer in effect.  The first step was already done in May of 2013.  If the pool becomes a valid 51% threat again, a new plan will be put forward.

This is being posted in a new thread so that it stands out to people who do not frequent the primary pool thread, since this is about more than just the pool.
A lot of noise in the IRC, reddit, and forum yesterday related to "BTC Guild dangerously close to 51%" due to a large amount of luck yesterday - https://slotsonlinecanada.ca/.  The pool found ~30% more blocks than expected at its given hash rate.  According to the last 2016 blocks, BTC Guild is still shy of 40% (36.61% as of this post).  Obviously I can't wait until the pool is 49.9% to start taking measures, even though a 51% attack is only a true threat if the person controlling it uses it.

This is the outline for measures that will be taken.  I will not be using 24-hour pie charts from blockchain.info to base these decisions due to how much luck influences the charts (either good luck by BTC Guild or bad luck on the rest of the network).  Figures will be pulled from http://blockorigin.pfoe.be/top.php which accurately grabs each block for BTC Guild, and also uses a 2016 block window to determine percentages.


If Pool Speed is Over 40% of Network
BTC Guild will begin limiting the creation of new accounts.  Additionally, the fee on PPS will be increased from 5% to 7.5% on all new miners, and will be moved to 7.5% on old miners after the difficulty changes.  PPLNS will remain at the 3% + tx fees rate initially.


If Pool Speed is Over 45% of Network
BTC Guild will remove all getwork based pool servers within 24 hours.  This is expected to reduce the pool by about 3.5 TH/s, or roughly 15% as of this post.


If Pool Speed is Over 45% of Network After Getwork is Removed
PPLNS fee will be raised to 4%, and new registrations will be completely closed off until speed drops back under 40%.



Suggestions are welcome if you can think of a better way to make miners willingly leave the pool.  The only thing I will not consider is kicking miners off entirely (outside of getwork).

[1714629368]

1714629368

[1714629368]

1714629368

[1714629368]

1714629368

[1714629368]

1714629368

[ruski]

Looks like I jumped in just at the right time then.

[alir]

Thanks for doing this, I'm glad we have plenty of level-headed people with the communities interest in mind.

[Chris Acheson]

Have you considered increasing your fees if your speed stays higher than your target?

[iCEBREAKER]

User # [<1000] here.

If anyone wants to buy my OG (original guildster) account send me a PM.   

[Gaff]

Why not just pay less per share? That way the economic incentives lines up with the result we'd like. win:win!

[eleuthria]

I've considered raising fees, but Guild already charges reasonable fees (3% and paid for orphans+tx fees or 5% straight PPS).  It's a good spot, and it's made me enough to be worth the 16+ hour work days when a crisis happens.

Raising fees is just penalizing users who stick around.  I might make more, I might make less, but it's at the cost of the users that don't leave.  It also leaves a bad taste in people's mouth when a pool raises fees.  Pools have always had a strong leader, and it's changed hands a few times in the last two years (Slush -> Deepbit -> BTC Guild -> Deepbit -> 50BTC -> BTC Guild).  I'm trying to come up with a solution that doesn't damage the pool long term if the lead changes again.  I don't want to be in the position of 51%, but I also don't want to cut off the legs of my business in the long run.

I've been a bit too defensive (which turns into aggressive) in IRC the last 24 hours about this, and I apologize to those that were on the receiving end when they raised the concerns.  This is something that has been on my mind a lot for the last month.  Everybody expected the network speed to explode once ASICs delivered.  But I don't think anybody expected a single pool to get such a disproportionate amount of the first batch.  The pool has increased to 6x more GH/s than what was there at the start of February.



EDIT/UPDATE:  After a lot of input (mostly in IRC), the above statement no longer applies to the new plan.  Quite simply, raising fees is the only effective way to make users consider choosing another pool and willingly leave.  I'm giving a large amount of warning now with this post, and it's also referenced on the pool website news page.  I hope it's not required, but it really is the only method that we've come up with to make users leave without arbitrarily kicking them off the servers.

[SgtSpike]

How about the PPS fee raises by 10% of the amount mined for each 1% above 45%?  In other words, at 50%, the users are paying a 53% fee to mine there?  That way, you're not penalizing the users until they refuse to switch over, and only then.

[eleuthria]

How about the PPS fee raises by 10% of the amount mined for each 1% above 45%?  In other words, at 50%, the users are paying a 53% fee to mine there?  That way, you're not penalizing the users until they refuse to switch over, and only then.

The biggest problem with raising fees is miners do not keep constant check on the website/forum thread.  If you have automatic payouts and self monitoring, there is very little reason to even look at the pool website.  Even less reason if you use the pool API.  Raising fees, in my opinion, requires significant warning time.  Users will feel cheated if they log in after a week and see the last few days they were being charged more than they signed up for.  Lowering fees is easy, nobody complains about extra money, but the other way around just leads to a lot of headaches.  The switch to PPS with BTC Guild had a lot of warning, and users still emailed me complaining -weeks- after the switch.

[iCEBREAKER]

What about splitting the pool up into two or more?

One for merged mining via getwork, one for GPU stratum, and one for (20+GH) ASIC stratum?

[willphase]

Obviously I can't wait until the pool is 49.9% to start taking measures, even though a 51% attack is only a true threat if the person controlling it uses it.

people will think you're being so reasonable and committed towards the Bitcoin project by putting in place these preemptive countermeasures that it will only attract more people to mine on BTCGuild.  You need to be more... more... BFLish, and push people away

Good job though!  I might mine on BTCGuild!  Oops I mean er.. I'll stay elsewhere... damnit!

Will

[nbtcminer]

why do all the above and close registrations until TH cools off?

[infoporter]

One way to easily mitigate this would be for someone to release some sort of generically configured Bitcoin Mining Pool application or packaged archive that is easy to install and get running to help grow the mining pool community.   Even with your mitigation efforts, the true problem lies in the fact that there are only a few large mining pools. The more other mining pools that spring up the better for the decentralization of the overall Bitcoin network.

[geckoman]

What about splitting the pool up into two or more?

One for merged mining via getwork, one for GPU stratum, and one for (20+GH) ASIC stratum?

That doesn't matter if all split off pools stay under the control of the same person or group. It would just make the problem less obvious to the general public.

[crazyates]

Are you talking about just shutting off the GW servers, or turning them into a proxy pool that would mine on other pools?

[eleuthria]

Are you talking about just shutting off the GW servers, or turning them into a proxy pool that would mine on other pools?

It would be turning them off.  I have no intentions of even attempting to run a proxy pool.

[S M I L Y]

How about an increasing graduated fee for newly registered users?    This will encourage new users to look else-ware for another pool.

Once BTCGuild's relative hash rate drops below X% you can then lower the fees for new users that decided to stick around and pay the higher fee.

This rewords your hard work with some extra coin and encourages a free market approach to controlling your hash rate.

[ldrgn]

How about you screw externalities, keep on adding users and let the free market sort it out?

[Luke-Jr]

How about you screw externalities, keep on adding users and let the free market sort it out?

This is part of the "free market sorting it out":
Having a single entity with blind control over even 25% of the network blocks is a threat that devalues everyone's bitcoins (including their own).
Therefore, it makes economic sense for BTCGuild to try to deter this from occurring.

Too bad they won't allow miners to audit their block data by supporting GBT.

[eleuthria]

I have updated the plan a little bit to proposals that include increased fees to push some miners out to other pools, rather than relying on registrations being cut off.  Cutting off registrations will not prevent old miners from turning on ASICs they receive, and with over 80,000 accounts on BTC Guild, it's far more likely for the new hash rate to come from old users than new ones at this time.