Jan 24

Bitcoin Mining Experiment End

I am going to end the Bitcoin mining experiment as it currently stands because it has proved what I needed it to prove. Namely it debunked several popular theories on bitcointalk.org and exposes exactly why BTCGuild is the most popular pool for users allowed to choose. Another reason for stopping it is because the block erupters by themselves are not going to be able to keep up with difficulty. I’m going to add them back to the farm.

I need to wait several hours for the remaining shifts on BTC Guild and ghash.io to clear, and also for the DGM rewards to fully pay out. As a preview:

BTCGuild earns more BTC than any other pool tested.

cex.io/ghash.io is not a “guaranteed loss” in fact you would have to be pretty stupid in order to lose money there by mining.


Jan 17

Good news on the NSA front?

From CNN

In a speech at the Justice Department today, President Obama will call for an end to the National Security Agency’s controversial bulk phone record collection program as it currently exists, a senior administration official told CNN Senior White House Correspondent Jim Acosta.

According to the official, the president will recommend that the collection of Americans’ phone records remain at the NSA temporarily as he seeks input from Congress and the U.S. intelligence community on where to store the data permanently.

The president will also take steps to modify the program so that a “judicial finding is required” before the NSA queries the database.

For complete coverage of breaking news, go to CNN TV, CNN.com and CNN Mobile.

I have the perfect place for it to be safe forever: Send it to /dev/null.

Unfortunately what I really think is going on here is just a diversion. They’ll just go back to claiming that they are not collecting records and everyone will go back to believing it, simply because that is easier to believe and it requires no responsibility of the citizen to actually do something about it.

Jan 11

Bitcoin Mining Experiment Four Week Status

It has now been 4 weeks since this experiment started. In a surprising turnaround, Ghash.IO is now break even with BTC Guild. Currently running with an aggregate hash rate of approximately 415 MH/s, it remains to be seen whether that benefit over the other pools is enough to overcome the perpetually falling GHS price.

There are two points I’d like to make at this point in the experiment. First is the return on investment. With the average value of 1 BTC being approximately $880 USD at the moment I write this, this makes the total USD value of a month of mining on a block erupter to be approximately $3.45. Now people are selling on eBay hash power for many times this value. This means one of two things, either that the buyers expect the price of BTC to go up by a factor of 10 or so, or they don’t know what they are buying. I suspect the latter. The second point I’d like to make is that CEX.IO is not a “guaranteed” loss as so many people on bitcointalk.org like to trumpet. It is only a guaranteed loss if you are too stupid or impatient with the market. This experiment started with a set of parameters, has followed those parameters precisely, and if it were to cash out now would be break even with BTC Guild. Anyone could have done this, it is not rocket science, and so anyone could have made a profit using CEX.IO. End of story.

As of 201401111120 (Eastern US)

  • BTC Guild: 0.00392178
  • Eligius: 0.00387029
  • EclipseMC: 0.00343288
  • Ghash.IO: 0.00392059 (0.08167889 GHS * 0.048)
Jan 09

Ghash.IO and the state of Bitcoin

There seem to be a lot of people making an issue about the increasing hash power of ghash.io at the moment, so I’m going to post my comments here. First of all, however I’d like to point out that ghash.io has already attempted to double spend bitcoins even without having the easy ability to do so. The times this has happened in the past, the network was not taken advantage of with the so-called 51% attack. Since ghash.io has already proven themselves, as can be seen in the blockchain, to have attempted to cheat the system, no one should have any faith that they will do the right thing if they do get to 51% of the network.

People are drawn to Ghash.IO primarily because it has a high hash rate which provides quick results (people like instant gratification). The website is slick with the automatically updating stats. It merge mines several coins (perceived value, even if some of those coins are not worth much). There is zero fee. It allows someone to automatically split up the mining proceeds by percentage, for example, for a group buy. There is also the gambling / trading aspect of cex.io which has its own lure. Additionally, and perhaps most hurtful of all is the marketing aspect that incentivises people to advertise for them.

Ghash is a big threat, and there is no way to know for certain that cex/ghash do not have additional hashpower running on other “unknown” private pools. Knowing that they will hurt themselves if they dump too much into cex.io at once. It is also unknown just what the ratio of their own hash power to individual’s personal hash power pointed at that pool is. The other threat is that there is no representative (that I have seen anyway) that interacts with the community. They do not seem to be publicly interested in cleaning up their name in many of the threads about their past with double spending, for example. Ask yourself for a moment – how do they make any money themselves? They charge no fees for trading, minimum fees for withdrawing, very minimal fees for their own hash rate rentals. There is something else going on there, and this puts them in a class by their own.

I do not agree that something “needs to be done” about BTC Guild. The pool admin seems a very standup guy who is himself interested in the health of the community. I think that if it ever did come down to short term personal gain or the community, he would do the right thing for the community, and ultimately his own long-term personal gain as well. The vast majority of the hash power there is voluntary – meaning that people could pull their miners en masse and the pool could instantly be smaller. The pool has also not had the same bursting growth rate as ghash.io, and right now is also the biggest weapon the community has in fighting cex.io if it attempted to run a 51% attack. When bringing up new mining equipment, putting it on any pool other than those two is probably the best thing any of us can do individually, but it doesn’t help that (from my testing) the most profitable pool is BTC Guild – in spite of its fees. If I’m in a group buy or trying to get a full return on a piece of hardware, it is silly to mine on a pool that I know will return a lower profit over enough time that I know variance isn’t the cause.

Anyway – my 2 cents. I think what the rest of the pools need (other than the top 2) is very clear, but I’m going to spell it out briefly below. Note that this is in addition to implementing the features that set GHash apart from the rest such as split payments, and slick stats:

Eligius – Transparency with the shelved shares and namecoin. BTC Guild has transparency and the ability to audit work that Eligius lacks.

Slush – An operator who cares about the pool. Slush is absentee and the pool suffers from payment and accounting issues. Why anyone still mines there is odd.

EclipseMC – This pool has the best implementation of DGM out there, however some people will never mine there with Josh from BFL at the helm. From my own testing, it also pays out less than BTC Guild and Eligius for the same amount of work. See the experiment link for more info.

Any other PPLNS pool: Too low value of N.

Any other pool has too low hashrate to keep variance at bay. I think some pools are going to have to take a loss temporarily or go PPS to some extent in order to pull hash rate away from ghash. Remember, however, that this is just covers the personal power pointed at ghash. We will not be pulling cex.io power away, although to the extent cex.io can be believed, that is a feature that they are planning to implement.

Dec 28

Bitcoin Mining Experiment Two Week Status

As of 201312282200 (Eastern US)

  • BTC Guild: 0.00227196
  • Eligius: 0.00211533
  • EclipseMC: 0.00205489
  • Ghash.IO: 0.00183122(0.03799221 GHS * 0.0482)

This project is now two weeks in and I am ready to make a statement of sorts. Had Ghash.IO GHS rate remained constant, it would have been the clear winner. As it is, it is so far the clear loser. However, the gain/loss is not realized until I cash out, so I am going to continue to hold and maintain this experiment over time to give Ghash.IO every opportunity to redeem itself either by providing an accelerated hash rate due to continuing reinvestments or by recovering the GHS price.

BTC Guild 0.00227196 100.00%
Eligius 0.00211533 93.11%
EclipseMC 0.00205489 90.45%
Ghash.IO 0.00183122 80.60%

The numbers may not make a lot of sense at the 333MHs rate, but blow these numbers up to be a Block Erupter blade setup at 107GHs and the rewards would play out as follows:

BTC Guild 0.72929916
Eligius 0.67902093
EclipseMC 0.65961969
Ghash.IO 0.58782162

That’s significant, especially when both EclipseMC and Eligius claim 0% fees and BTC Guild 3%. Before anyone cries luck, I’d like to mention that this was a 2 week experiment, which is the amount of time planned for difficulty adjustments. If you made an investment into ASIC hardware, breaking even becomes a priority over many other issues including stats, network diversity, owner likability, etc. The danger right now is this: BTC Guild is, in my opinion, winning the game in almost every respect listed including reward. The only way to keep the network diverse is to have someone develop a competing pool that can match the reward of BTC Guild.

Dec 24

Bitcoin Mining Experiment Status as of 201312240900

As of 2013240900 (Eastern US)

  • BTC Guild: 0.00165256
  • Eligius: 0.00160620
  • EclipseMC: 0.00147493
  • Ghash.IO: 0.00153935 (0.02552822 GHS * 0.0603)

Eligius implemented an estimated daily earnings, which is a really nice piece of information:

Estimated Earnings
Your approximate maximum potential earnings at the current network difficulty of 1,180,923,195.26 and maintaining your 3-hour average hash rate of 311.78 Mh/s is 0.00013278 BTC per day.

Ghash.IO/Cex.IO price per GHS is significantly down now from when the experiment began, however its payout should I cash out now is still higher than EclipseMC.

Dec 23

An Experiment in Bitcoin Mining – Part 7 – BFL Strikes!

I had hoped to start this experiment with ButterflyLabs (BFL) equipment, but started instead with Block Erupters. While I think there will be little difference besides magnitude between the devices, I had hoped to show that it was possible to get some kind of a payout from a pool with this. It appears that the block erupters may never see a payout, so I was still hoping to add the BFL equipment when it arrived.

The equipment arrived today, however during preliminary testing I have found that it is anything but consistent. The following is the hash rate of these four devices:

2013-12-23 17_22_39-guardian - PuTTY

Several things wrong with this picture. First, the original BFL device I got has never had a hardware error. These are already racking up the errors similar to block erupters. Second, they are all hashing at different rates. It will be impossible to dedicate one to each pool as I have done with the erupters.

Several thoughts I had. First, I could attempt to use the balancing function of the mining software to spread blocks evenly among the pools, or I could do round robin, or I could rotate them all among the pools. The problem with any of these, except load balancing is that it would potentially increase the error factor of luck. With load balancing I would have to start keeping track of the individual client details to ensure proper operation. I am not sure any of them really do load balancing all that well.

The third thing I could do is calculate a handicap based on each one’s hashing power. While I think this would work just fine for straight pools, it wouldn’t reflect the amortizing power (or penalty, depending how you view it) of cex.io/ghash.io.

Another idea would be to use additional block erupters to get all the systems within < 333 MHs of each other. For example, I would need about 7 block erupters to boost the power of #1 above to the power of #3. At this point a smaller handicap could be calculated, and there would be less impact on the cex.io compounding.

Thoughts/Comments? I’m going to think on this a while and allow these to burn in with my other miners not affiliated with this experiment.


Dec 23

Bitcoin Mining Experiment Status as of 2013231045

  • BTC Guild: 0.00154781
  • Eligius: 0.00140232
  • EclipseMC: 0.00141193
  • Ghash.IO: 0.0015161 (0.02307720 GHS * 0.0657)

Ghash.IO/CEX.IO GHS rate continues to decrease, however they remain in second place. Had the GHS cost remained consistent where this experiment started at 0.0735, they would now be at 0.00169617 and clearly in the lead. Since the price fluctuates wildly at times, a smart trader would certainly not sell at a low.


Dec 19

Bitcoin Mining Experiment Status 20131219

As of 2013191900 (Eastern US)

  • BTC Guild: 0.00102693
  • Eligius: 0.00088874
  • EclipseMC: 0.00095393
  • Ghash.IO: 0.00096181 (0.01375989 GHS * 0.0699)

Before I begin commenting, let me say first that all the variance between these scores may be due to the luck of the respective pools, and we may swing back around. I am looking for long-term patterns here, so real evidence of any pattern is unlikely to show up for at least a month, unless there really is something wrong with a pool’s payout mechanism in relation to another pool’s.

The first thing that jumps out is that Ghash.IO/CEX.IO pair is flying in the face of conventional wisdom, at least wisdom on bitcointalk, which probably isn’t saying much. Conventional wisdom there says that the platform is a sure way to lose money. I have argued a few times that only people buying on hype and panic selling are going to see huge losses. What I failed to do was take into account my audience. Arguing with an idiot just brings a person down to their level. Truthfully, there are not a lot of idiots on bitcointalk, but there are a lot of young people who have very little experience or knowledge outside of their observed truths.

The second thing jumping out at me is the Eligius reward. I like Eligius for several reasons, but I always thought my reward was lower while I was mining there. Now I have some proof. Imagine if I had a chili miner instead of a block erupter. My hash rate would be 100 times, and presumably so would be my reward at this point in the experiment. See the following:

  • BTC Guild: 0.10269300
  • Eligius: 0.08887400
  • EclipseMC: 0.09539300

That is a significant difference between the top and bottom place. I have not included Ghash.IO in this extrapolation because it wouldn’t scale quite the same way due to the income reinvestment. By this point it would have more than an extra 1 GHs of power on its side, so like compounding interest, it would be doing better than what I could show with just a linear * 100.

As I said, this is likely just luck. As you can see from the graph of Eligius, the maximum reward (green area) shows that my shares are being shelved.2013-12-19 19_11_37-Clipboard

In effect, what this is saying is that if I had mined at a PPS pool I would have the green line, minus any fee for PPS. From past experience I know this works its way out eventually, even if I stop mining at the pool. Still, this kind of temporary loss of reward might be difficult for some people to comprehend. There really should be some sort of way to better visualize the shelved shares and a guess of some sort as to when they might be paid. Just my thought here.

EclipseMC has had some absolutely horrid luck the past few days. Being able to maintain placement against BTC Guild which has had good luck is also telling. Again, I look forward to being able to report similarly as time progresses.


Dec 17

Bitcoin Mining Experiment Status 20131217

As of 2013172300 (Eastern US)

  • BTC Guild: 0.00061516
  • Eligius: 0.00057260
  • EclipseMC: 0.00068595
  • Ghash.IO: 0.00063825 (0.00880357 GHS * 0.0725)

I find it interesting that this result already matches my perceived strength and expectation out of each pool. I realize that it is still way too early to draw any conclusions due to luck.

I will go ahead and share some additional biases I have about Eligius. Just reading the payout method makes me think that the following. There is double negative impact from poor luck. First of all, rewards confirm slower so that the miner receives less coin per set of time. Second, shares shelve which may or may not be paid back in subsequent lucky rounds. There is single or -maybe- double impact from positive luck. In either case, one has to mine through unlucky rounds before lucky rounds pay the second part of the benefit. The best you can hope for is to hit a normal PPS rate. The worst is that you are helping pay other people who have mined the pool longer and have more shelved shares.

My favorite, EclipseMC gets that vote because shares are never lost to long shifts. In BTC Guild this rarely happens because the value of N including the shifts is so large, but with the shorter value of N in ghash.io, one could point significant power at the pool for, say, ten minutes and then stop, and if it takes longer than an hour to find the next block that hashing power would have absolutely no positive benefit.

What I really wish is that EclipseMC had a hashrate at least the size of Eligius.

We shall continue to see if this theory holds up and EclipseMC can stay ahead. The real wildcard here is the ghash.io/cex.io account. While it may be doing great at the moment, it just as easily could lose value. As tempted as I am to play the market, I will refrain.