What (if any) mechanism is there to protect against a massive hash rate drop? - page 2.

coinft

full member

Activity: 187

Merit: 100

Quote from: pyra-proxy on April 23, 2013, 06:13:52 AM

And how is that easier than upsetting the difficulty repeatedly on purpose over a span of Months vs. a span of Days/Hours before having to come back and repeat the task again?.... This is effectively what is happening to Terracoin right now, and I bet if you ask most users, would you prefer the namecoin experience (where it was taking many months to have a difficulty drop, and block times that were numbering many hours) or the Terracoin experience (where, ya hashing could hop in and out repeating in a shorter time than before but at least difficulty fixes itself fairly rapidly when the disproportionate hash rate comes and goes)

What I'm trying to say is that upsetting difficulty is easy both ways, the "Terracoin" way though leaves the end user with a workable network, whereas the current bitcoin way leaves the end user with a horrible system to operate on for a long time.... I would think the latter is a lot more damaging to the adoption rate.

Terracoin is weak and desperate and needs to defend against the top dogs somehow, even if it has drawbacks. The best defense of bitcoin is being the ultimate top dog.

TierNolan

legendary

Activity: 1232

Merit: 1094

A simple rule is build on the valid chain with highest POW.

A chain is valid if
- the head of the chain meets the standard difficulty
- the head of the chain was seen by the miner at least f(block difficulty) minutes ago

Re-targets would be every 2016 blocks, but would take total POW between the 2 endpoints as the target.

The function could be a decaying exponential. For example, the target could drop by 50% every 10 minutes, after the first 20.

f(block difficulty) = 20 + 10 * log2( (standard difficulty) / (block difficulty) )

Blocks which meet the standard difficulty are zero.

If the block has 1/64 of the difficulty, then it would be forwarded to peers, and then held in a queue for 20 minutes + 60 minutes. In the normal situation, another block would arrive that meets the standard difficulty and thus that block would be discarded.

pyra-proxy

hero member

Activity: 490

Merit: 500

Quote from: Etlase2 on April 23, 2013, 12:49:59 AM

Quote from: Impaler on April 23, 2013, 12:05:16 AM

A simple decrease in the number of blocks between difficulty changes would go a long was to address this as an issue. In BTC it is 2016 which is 2 weeks at target rate. This is far longer then necessary to smooth the variance in finding times and makes difficulty changes unnecessarily sluggish and vulnerable to rate crashes.

Don't forget that decreasing the number of blocks between difficulty changes would also make it easier to upset the difficulty repeatedly and on purpose. Simple fixes tend to have unintended consequences.

And how is that easier than upsetting the difficulty repeatedly on purpose over a span of Months vs. a span of Days/Hours before having to come back and repeat the task again?.... This is effectively what is happening to Terracoin right now, and I bet if you ask most users, would you prefer the namecoin experience (where it was taking many months to have a difficulty drop, and block times that were numbering many hours) or the Terracoin experience (where, ya hashing could hop in and out repeating in a shorter time than before but at least difficulty fixes itself fairly rapidly when the disproportionate hash rate comes and goes)

What I'm trying to say is that upsetting difficulty is easy both ways, the "Terracoin" way though leaves the end user with a workable network, whereas the current bitcoin way leaves the end user with a horrible system to operate on for a long time.... I would think the latter is a lot more damaging to the adoption rate.

Etlase2

hero member

Activity: 798

Merit: 1000

Quote from: Impaler on April 23, 2013, 12:05:16 AM

A simple decrease in the number of blocks between difficulty changes would go a long was to address this as an issue. In BTC it is 2016 which is 2 weeks at target rate. This is far longer then necessary to smooth the variance in finding times and makes difficulty changes unnecessarily sluggish and vulnerable to rate crashes.

Don't forget that decreasing the number of blocks between difficulty changes would also make it easier to upset the difficulty repeatedly and on purpose. Simple fixes tend to have unintended consequences.

pyra-proxy

hero member

Activity: 490

Merit: 500

Quote from: Impaler on April 23, 2013, 12:05:16 AM

A simple decrease in the number of blocks between difficulty changes would go a long was to address this as an issue. In BTC it is 2016 which is 2 weeks at target rate. This is far longer then necessary to smooth the variance in finding times and makes difficulty changes unnecessarily sluggish and vulnerable to rate crashes.

This was 1 part of how TRC fixed this problem for them... they've had asics coming in and out for days now, difficulty ramps up pretty quick and then when they bail difficulty comes down to normal miner level pretty quick. The other part of their solution involved how they determine the next difficulty. IMHO bitcoin would be better off rolling this algo into it and start reaping some reward from the numerous alts that were inspired by it.....

RandyFolds

sr. member

Activity: 448

Merit: 250

Greed. People were still mining when it dropped from $33 to $2.

Impaler

sr. member

Activity: 826

Merit: 250

CryptoTalk.Org - Get Paid for every Post!

A simple decrease in the number of blocks between difficulty changes would go a long was to address this as an issue. In BTC it is 2016 which is 2 weeks at target rate. This is far longer then necessary to smooth the variance in finding times and makes difficulty changes unnecessarily sluggish and vulnerable to rate crashes.

kjj

legendary

Activity: 1302

Merit: 1026

There is no such mechanism. Furthermore, any possible mechanism would be worse than the (alleged) problem it hopes to solve.

Feel free to search out any of the many, many threads on this topic if you'd like more details.

qxzn

hero member

Activity: 609

Merit: 506

Quote from: kaerf on April 22, 2013, 05:25:51 PM

If you're unaware, Terracoin had a problem a couple weeks ago with ASICs shooting up their diff then leaving the network (thus leaving the network at a very high difficulty relative to remaining hashing power). I think NMC had the same problem and it was resolved with merged mining, but that effectively killed NMC(??).

I wanted to ask what mechanisms in BTC there are (or can be) to prevent a massive hash rate drop off. Right now, we have some massive amounts of hashing power being generated by Avalon and ASICMINER. If there was a bug in their firmware (e.g. the recent large coinbase bug that dropped all avalons on eligius to 0 MHS) that caused all Avalons or AMs to drop off the network temporarily or permanently, it could take days for a new block to be found. This would wreak havoc on the entire bitcoin ecosystem by bringing transactions to a standstill.

Of course, as more manufacturers produce ASICs this problem is reduced, but let's say BFL delivers and their customers end up being 30% of the network...we'd still have the same problem. (We're all very confident in BFL's ability to produce a quality product Grin

)

Is this a legit concern? Shoot me down. I'll feel better.

I believe there is no such mechanism, and it is a legit concern.

Remember people, bitcoin is an experiment

. Don't risk your life savings!

rarkenin

hero member

Activity: 784

Merit: 500

By the time there is a massive increase in difficulty, there will most likely be different ASIC manufacturers, so one mass failure would affect a smaller portion of the hashing power. This portion could still be on the order of 80%, though.

kaerf

hero member

Activity: 631

Merit: 500

If you're unaware, Terracoin had a problem a couple weeks ago with ASICs shooting up their diff then leaving the network (thus leaving the network at a very high difficulty relative to remaining hashing power). I think NMC had the same problem and it was resolved with merged mining, but that effectively killed NMC(??).

I wanted to ask what mechanisms in BTC there are (or can be) to prevent a massive hash rate drop off. Right now, we have some massive amounts of hashing power being generated by Avalon and ASICMINER. If there was a bug in their firmware (e.g. the recent large coinbase bug that dropped all avalons on eligius to 0 MHS) that caused all Avalons or AMs to drop off the network temporarily or permanently, it could take days for a new block to be found. This would wreak havoc on the entire bitcoin ecosystem by bringing transactions to a standstill.

Of course, as more manufacturers produce ASICs this problem is reduced, but let's say BFL delivers and their customers end up being 30% of the network...we'd still have the same problem. (We're all very confident in BFL's ability to produce a quality product Grin

)

Is this a legit concern? Shoot me down. I'll feel better.

Topic: What (if any) mechanism is there to protect against a massive hash rate drop? - page 2. (Read 1894 times)