Yes. ThickAsThieves posted Friedcat's response here:

https://bitcointalk.org/index.php?topic=99497.msg2460719#msg2460719

Reminder -- We're collecting questions from the community that will be compiled and voted on (by the community as well,) and passed on to Friedcat by TAT.

If you have questions that you feel are important, please submit them via PM to either ThickAsThieves, FrankTank, VJain or myself.

Ian

Sure.

Will they? No. I wouldn't if I were them. It is definitely a competitive advantage to keep that information proprietary.

For all we know they could be cycling through new hardware at this very moment. If they're cycling through 40 th/s groups of hardware for testing purposes and they left three groups on at once.. Competitors would know "they have at LEAST 120 th/s of hashing power available at any moment."

Whereas if they do not do that, and we infer that there was 100+ th/s over the previous hour, that can be attributed to variance.

Well, not bogus, but not reliable. Anything with a 4 unit sample size is error prone.

However, you can make some pretty interesting assumptions off of the data.

I posted yesterday that when I see a series of blocks mined that looks like "AM-AM-BTCGUILD-AM-AM", that I know it's variance (or luck.)

However, if I saw the same thing that looked like "AM-AM-BTCGUILD-AM-AM-AM" then I would believe that it is due to a higher quantity of hashing power than is reflected in the 24 hour average.

There's no real way to quantify what I'm explaining, it's just a feeling I get from looking at the data. Right or wrong, it's still interesting to look at.

Now -- with a 48 hour sample size -- we *should* get a pretty good estimate of what the actual average hashing power is over that time.

Regardless of who that statement was intended towards (I think it was intended towards me,) take it as an opportunity to learn more about how hashrate is derived at the current difficulty. It's interesting stuff

The answer to that is a definitive no.

Most people do not realize that we do not have actual hashrate data from ASICMINER, only inferred hashrate data that we figure based on number of blocks found during a given timeframe and the current difficulty.

Again -- you have to know the sample size that he's using.

http://www.dpcapital.net/blockchain/?hours=24,48,72,96,120

If you look at that, you can see the 24 hour sample size matches exactly to that of runeks.dk/bitcoin.

In other words, at the current difficulty, it will take 46 th/s to find two blocks in one hour.

Just so everyone knows, all of the ASICMINER data that is presented is inferred. Because they (ASICMINER) do not post the actual hashrate, we have to infer the hashrate from the number of blocks found in a given timeframe and the difficulty.

If we did know the hashrate, we could do fun things like determine the probability of finding 5 blocks in an hour at the current difficulty.

They all come from the same source -- blockchain's json API.

http://www.dpcapital.net/blockchain/?hours=1,2,4,8

You can change the querystring "hours" parameter if you want to look at customized views.

He made it clear -- in 1 week he will be ready to sell. duhhhhhhh

I dunno.. That's a lot of variance.

It is possible, but we have no way to determine what the probability is unless we know the *actual* hashrate.

On the verge of another hashrate drop?

http://www.dpcapital.net/blockchain/?hours=1,2,4,8

Get the backing of some big-time shareholders and go for it. Also offer escrow service.

Coincidence. I saw one earlier with 600+ transactions.

Regardless, there is no mechanism in BTC to selectively mine blocks. I.e., you can't mine a 100 transaction block more easily than you can mine a 1,000 transaction block with a given difficulty. This cannot be done by adjusting hashrate either. At a given difficulty, your probability for mining a block is fixed to a certain hashrate over a period of time. A lower hashrate will have a lower probability, and a higher hashrate will have a higher probability.

In short, hashrate and difficulty have nothing to do with the size of the block you mine.

Ian

No problem -- For what it's worth, I have a strong background in statistics and probability. But some things are just too difficult to quantify, and or we don't really have enough information to quantify.

I think with ASICMINER this week, we're the later.

Ian

I'll tell you my personal take -- when you see a streak of 2 or 2, skip one, and 2 more.. Anything after that is typically not margin of error/luck/fluke. For example, 2-x-2-x-1 would probably be due to a lot of hashing power. If you see a 2 or 2-x-2 streak, then it's likely luck. With that said, I would speculate that the recent 2-x-2 finds are likely luck. I would attribute possibly 1 of those to the hashrate, and 3 of them to luck.

Just an observation that I can't quantify.

Margin of error in 24 hours is fairly low, while 48 hours is half of 48 hours. Each one of the sites that shows you a graph takes an hourly hashrate and projects from there. 24 hours has 24 separate samples, 48 hours has 48, etc.

I'm a businessman by day (and night,) but my personal opinion is that jokes keep things light hearted. Nobody should take this too seriously. After all is said and done, we still have families, friends, etc.

I can see how this would be viewed as spam'ish, or rhetoric by someone who doesn't feel the same way, though

Ian

FYI -- you can change the hours parameter on the query string to view custom charts. Default is the 4 day chart. Sometimes it's handy to look at 1,2,4,8, etc.

Data is available as far back as late last Thursday (6-13) up to what is current from the blockchain API.

Ian

Down to 69.22 th/s in the last hour. PANIC SELL

The probability of this happening at the time I posted is exactly 100%

Ian