Don't expect another 8% Diff increase -- expect more

[AngelusWebDesign]

FYI -- the next difficulty increase will be larger than the last one (8%); maybe even larger than the one before that (13%). So can we still say difficulty is leveling off, or has leveled off -- or is it rising at a more substantial rate again?

Today we just jumped to 16 Thash/s.

I know, everyone just got lucky; we really are holding steady at 11.5 Thash/s, and it's all just a glitch or a streak of awesome luck for everyone at once. Don't worry, not a single 5770 was actually added to the network. 

But it does seem that capacity is being added. Because if it were just luck, it would go down just as often (statistically), and there would be 0 difficulty increase.

We're rising faster than during the last difficulty increase -- so it's as I expected. You have a "big boom" (early June, when BTC was $30), followed by an "echo boom" after difficulty starts leveling off and people shout "All clear!" as they run to the store and fill their cart with 6870's.

As many have predicted, BTC is going to have ups & downs (price, difficulty, network hashrate, you name it) as the system seeks equilibrium.

[1714008102]

1714008102

[1714008102]

1714008102

[1714008102]

1714008102

[grod]

Mining briefly had a completely insane ROI causing this instability in the first place.  7 day to get 100% return on investment while underlying assets generating that also appreciate?  Obviously that was unsustainable.

We'll stop having increases in mining power once mining is done exclusively on dedicated ASICs, with a below 10%+inflation annual ROI and bitcoins used as a stable currency.  Until then expect mining booms of various sizes driven purely by speculation -- just like $-to-bitcoin value. 

TL;DR Using anything less than 50% monthly difficulty growth for your buy-vs-mine bitcoins ROI calculations remains unwise at current $-to-bitcoin pricing.

[mike678]

Today we just jumped to 16 Thash/s.

I know, everyone just got lucky; we really are holding steady at 11.5 Thash/s, and it's all just a glitch or a streak of awesome luck for everyone at once. Don't worry, not a single 5770 was actually added to the network. 

You do realize people said the exact same thing before the last difficulty change. I can't remember what it specifically showed but it was close to 16Thash then as well. You need to just relax dude the difficulty is going to keep going up but this is definitely just luck again.

[kiwiasian]

Mining briefly had a completely insane ROI causing this instability in the first place.  7 day to get 100% return on investment while underlying assets generating that also appreciate?  Obviously that was unsustainable.

We'll stop having increases in mining power once mining is done exclusively on dedicated ASICs, with a below 10%+inflation annual ROI and bitcoins used as a stable currency.  Until then expect mining booms of various sizes driven purely by speculation -- just like $-to-bitcoin value. 

TL;DR Using anything less than 50% monthly difficulty growth for your buy-vs-mine bitcoins ROI calculations remains unwise at current $-to-bitcoin pricing.

In short, bitcoin is not worth enough

[AngelusWebDesign]

Actually, I've been charting the difficulty the past several sessions.

5 days into the last difficulty we maxed out at 13.4 TH/s, with about 7.2 blocks per hour.

We'll see if it calms down --

But all I said was it's going to be more than 8% or even 13% -- it could be 15%. Not exactly fire and brimstone.

I'm just making a prediction, that's all. I want it to be on record.

[mike678]

Actually, I've been charting the difficulty the past several sessions.

5 days into the last difficulty we maxed out at 13.4 TH/s, with about 7.2 blocks per hour.

its well over 16thash now, dot bit is showing a "instant" difficulty of 2.2xx.xxx, this does not seems to be a peak.

[1.21gigawatts]

Anusgelus is the oracle, listen to him.  Get out of mining.  The sky is falling.  Don't buy anymore video cards teenage fools

[AngelusWebDesign]

Actually, I've been charting the difficulty the past several sessions.

5 days into the last difficulty we maxed out at 13.4 TH/s, with about 7.2 blocks per hour.

its well over 16thash now, dot bit is showing a "instant" difficulty of 2.2xx.xxx, this does not seems to be a peak.

That second quote (shivansps) was from 7 days into the session, not 5. Difficulty reset on 7/6.

I never said anything about 7 days into the session. Last time, 5 days in, we didn't see much growth.

[mike678]

Actually, I've been charting the difficulty the past several sessions.

5 days into the last difficulty we maxed out at 13.4 TH/s, with about 7.2 blocks per hour.

its well over 16thash now, dot bit is showing a "instant" difficulty of 2.2xx.xxx, this does not seems to be a peak.

That second quote (shivansps) was from 7 days into the session, not 5. Difficulty reset on 7/6.

I never said anything about 7 days into the session. Last time, 5 days in, we didn't see much growth.

You do realize though that my original point is still valid that its just luck. Who cares if it was 5 or 7 days into the same period. Stop making these stupid fucking posts.

Yes the difficulty is increasing but this is luck.

[mike678]

I have reached a similar conclusion.

https://forum.bitcoin.org/index.php?topic=13339.msg380800#msg380800

The problem with your "predictions" is that you give your self such a large range so you cant really be wrong. Your low end for block 139104 is 3.5% and your high end is 31.4%.

[mike678]

I have reached a similar conclusion.

https://forum.bitcoin.org/index.php?topic=13339.msg380800#msg380800

The problem with your "predictions" is that you give your self such a large range so you cant really be wrong. Your low end for block 139104 is 3.5% and your high end is 31.4%.

The answer to that criticism is here:

https://forum.bitcoin.org/index.php?topic=13339.msg384337#msg384337

That 'range' as you call it is a probability distribution which indicates the amount of error, and the probability of that error as tested over the available history. The average deviation from the median forecast has been a mere 5.4%

If you could please inform me of a methodology that performs better I will happily employ it.

But the point is this. My methodology for forecasting the next re-target from pricing history has proven very reliable. And it happens to be in agreement with what is being asserted by OP.

There honestly isn't any better way to predict it unless you can see the future but all but two of your predictions have been around 6% off or higher. His prediction was 13% so it could very easily be 7-8%.

[bcpokey]

I have reached a similar conclusion.

https://forum.bitcoin.org/index.php?topic=13339.msg380800#msg380800

The problem with your "predictions" is that you give your self such a large range so you cant really be wrong. Your low end for block 139104 is 3.5% and your high end is 31.4%.

The answer to that criticism is here:

https://forum.bitcoin.org/index.php?topic=13339.msg384337#msg384337

That 'range' as you call it is a probability distribution which indicates the amount of error, and the probability of that error as tested over the available history. The average deviation from the median forecast has been a mere 5.4%

If you could please inform me of a methodology that performs better I will happily employ it.

But the point is this. My methodology for forecasting the next re-target from pricing history has proven very reliable. And it happens to be in agreement with what is being asserted by OP.

It's proven reliable because it's very easy. Most people do it without any mathematical modelling and can come within 10%. I disagree with the OP and his > 15% prediction, and we'll see who ends up being right (I've proven more successful thus far, but no one is perfect), I've done better than most just based on observation. The real difficulty lies in predicting farther out into retargets even farther down the road.

For example:

Block Number:   137088
  
Difficulty  
Lower Extrema:   1831645
Lower Quartile:   2153150
Median:   2851755

Crazy ass far off. Even the lower extrema wasn't really close. The previous one did a tiny bit better, but not much. When your error bars are almost 100% though that is hardly a statistically valid prediction, is what I assume that other fellow meant to say (especially when it is still wrong).

[mike678]

It's proven reliable because it's very easy. Most people do it without any mathematical modelling and can come within 10%. I disagree with the OP and his > 15% prediction, and we'll see who ends up being right (I've proven more successful thus far, but no one is perfect), I've done better than most just based on observation. The real difficulty lies in predicting farther out into retargets even farther down the road.

For example:

Block Number:   137088
  
Difficulty  
Lower Extrema:   1831645
Lower Quartile:   2153150
Median:   2851755

Crazy ass far off. Even the lower extrema wasn't really close. The previous one did a tiny bit better, but not much. When your error bars are almost 100% though that is hardly a statistically valid prediction, is what I assume that other fellow meant to say (especially when it is still wrong).

If you call 6.3% crazy far off then I suppose you are right.

But the figure you quote is for a different case, which was a forecast four re-targets out, I don't yet consider forecasts out that far to be reliable... Try and keep up.

But even so, if you know of a method that could do better for a Difficulty forecast that is reliable four re-targets out I would sure like to know what it is. So far nobody is stepping up to that challenge—I continue to work on it.

Like he said previously it's very easy to predict one or two difficulties out that's why you have for the most part been accurate. I'm pretty sure I could be within 6% if I guessed only 8% increases for the next two difficulty changes and then I would be just as good as you.

[grod]

Also, y'all realize 50% of the mining pool has had a 35% stroke of bonus luck in the past 24 hours, right?  Deepbit is running at 1.2 million shares to solve a block instead of 1.6M.  There's your "4 terahash of capacity" out of nowhere.

BTC is very much luck based.   That's why the "computing power" graph shoots all over the place.  With a target of 144 blocks solved per day you could have "good" luck or "bad" luck on even a dozen blocks and look like hashing power is imploding or exploding.

That's not to say the network isn't growing -- I'd be shocked to see less than 10% growth per 2016 blocks.  It's mostly throttled by the lack of available 5 series cards, not by lack of demand.  With new bitcoin community members joining at a steady clip some will choose to buy, some will choose to mine.  End result: network growth and a slightly smaller piece of the mining pie for all.  

Not the end of the world, I'll be happy generating .001 BTC/month per gigahash (Moore's law adjusted, of course) so long as that .001 is still worth $100.

[bcpokey]

It's proven reliable because it's very easy. Most people do it without any mathematical modelling and can come within 10%. I disagree with the OP and his > 15% prediction, and we'll see who ends up being right (I've proven more successful thus far, but no one is perfect), I've done better than most just based on observation. The real difficulty lies in predicting farther out into retargets even farther down the road.

For example:

Block Number:   137088
  
Difficulty  
Lower Extrema:   1831645
Lower Quartile:   2153150
Median:   2851755

Crazy ass far off. Even the lower extrema wasn't really close. The previous one did a tiny bit better, but not much. When your error bars are almost 100% though that is hardly a statistically valid prediction, is what I assume that other fellow meant to say (especially when it is still wrong).

If you call 6.3% crazy far off then I suppose you are right.

But the figure you quote is for a different case, which was a forecast four re-targets out, I don't yet consider forecasts out that far to be reliable... Try and keep up.

But even so, if you know of a method that could do better for a Difficulty forecast that is reliable four re-targets out I would sure like to know what it is. So far nobody is stepping up to that challenge—I continue to work on it.

Perhaps you should try and keep up with the statements I made in my short and concise post. I said next target is quite easy to do without any modelling of any kind. I said that the trick is making a model that can forecast farther down the road, and used your forecast as an example of how difficult it can be (Your median was about 70% off actual, and even your lower bound was wrong with a huge error bar).

I applaud your work on creating a method of prediction, but until you stop patting yourself on the back for being right with the easy stuff, I suspect you will have trouble with the hard stuff. An 8% increase still seems much more likely than a > 15% increase. So we'll see, you are predicting a roughly 17% increase for your next median I believe? Let us see who does better, just for fun, no wagers necessary.

Your mathematical model says 17.5% for median, I will say 8.5% based solely on my own observation and "gut feeling". Let us see who does better for the next retarget. I know you argue against hard target exactitude, but that is not the point of this game, merely to show the ease of next retarget checking. And I haven't even really been paying that much attention lately tbqh.

EDIT: 1834633 to be more clear.

[Kermee]

I know, everyone just got lucky; we really are holding steady at 11.5 Thash/s, and it's all just a glitch or a streak of awesome luck for everyone at once.

Average shares per block: 1524125 (-10.9%)
Average in last 24 hours: 1195922 (-41.4%)

I'll let someone do the statistical math of how that affects the 'derived' total Network Hash speed when nearly half of the total network hash pool is 41.4% lucky over the past 24 hours...

I'm sticking to my guns the next difficulty jump will be at or around 8% again, or less after the variance dies down.

Cheers,
Kermee

[Bitcoin_Silver_Supply]

As maaany have predicted... The BTC is going to have ups & dooowns... (price, difficulty, network hashrate, you name it) as the system seeks eeeequilibrium.

Hey everyone! An old man is talking!

[molecular]

I know, everyone just got lucky; we really are holding steady at 11.5 Thash/s, and it's all just a glitch or a streak of awesome luck for everyone at once. Don't worry, not a single 5770 was actually added to the network. 

On another note: a lot of mining power shifted back from namecoin-mining, which was twice as lucrative compared to bitcoin-mining. (see: http://dot-bit.org/tools/nextDifficulty.php), after the namecoin difficulty jumped from 23.000 to 94.000 on 23/07/2011 20:14.

[bcpokey]

I know, everyone just got lucky; we really are holding steady at 11.5 Thash/s, and it's all just a glitch or a streak of awesome luck for everyone at once. Don't worry, not a single 5770 was actually added to the network. 

On another note: a lot of mining power shifted back from namecoin-mining, which was twice as lucrative compared to bitcoin-mining. (see: http://dot-bit.org/tools/nextDifficulty.php), after the namecoin difficulty jumped from 23.000 to 94.000 on 23/07/2011 20:14.

I think people overestimate how much mining power was on namecoin. Haven't crunched the numbers myself, but I expect no more than a single terahash. Probably less.

[molecular]

I know, everyone just got lucky; we really are holding steady at 11.5 Thash/s, and it's all just a glitch or a streak of awesome luck for everyone at once. Don't worry, not a single 5770 was actually added to the network. 

On another note: a lot of mining power shifted back from namecoin-mining, which was twice as lucrative compared to bitcoin-mining. (see: http://dot-bit.org/tools/nextDifficulty.php), after the namecoin difficulty jumped from 23.000 to 94.000 on 23/07/2011 20:14.

I think people overestimate how much mining power was on namecoin. Haven't crunched the numbers myself, but I expect no more than a single terahash. Probably less.

I think about half a terahash during the "party". Thats roughly 5% of bitcoin power? Maybe just a dent, but when talking about 8% or 14% increase I think we shouldn't neglect it.