Topic: [POOL][Scrypt][Scrypt-N][X11] Profit switching pool - wafflepool.com - page 285. (Read 465769 times)

I neglected to mention one of my points in my earlier message.

Human beings do not always respond to a logical argument in a rational manner, and sometimes you have to bend a bit in order to trick them into seeing things your way.  Ultimately, your gross profits will increase if you attract more miners, and increasing pool hashpower will also benefit all miners too by reducing their variance.

The only reason doge is a sticking point is because it has the greatest capacity to skew short term results if we have bad luck with it.  And it is these immediate profitability numbers that matter so much in attracting more miners to reduce the effects of variance in the pool.  So it is essentially the whole chicken and egg situation going on.

And good luck while mining doge does not necessarily help either, as then expectations are set artificially high so when results regress lower towards the mean, some miners will leave because they perceive a decrease in profitability, unable to accept that the prior higher values were an effect of good luck.  So high variance coins like doge can be a lose/lose situation if building up pool hashpower is important to you.

But in two weeks at 100,000 blocks, perhaps this entire doge problem solves itself!

This.

And at the risk of belaboring the point, please add MEOW. It is a low difficulty coin that is consistently profitable.

Also TIPS. That's on cryptsy but only as a TIPS/LTC market. The automated trading should be able to handle that though, since it can already trade from LTC to BTC, right?

The more of these profitable low-diff coins get added, the less time will be spent on DOGE (with its inherent high difficulty variance risk).

Perhaps but c'est la vie.

Meanwhile there will be other more patient and intelligent miners flocking to the pool. I'm increasing my hash rate at this very moment.


That's a little different. I dropped off for example because it seemed that "something was wrong" (though I didn't know what).

I'm certainly sympathetic to the pool operator who wants to maximize his fees but for the rest of us it is actually better that the pool stay small so we don't drive up the difficulties (and drive down the profitability) on all the good coins.

I was not referring to wafflepool difficulties, but that other place that forces me to waste 5-10% of my electricity for no good reason.  Wafflepool is just fine in this regard and when you implement minimum vardiff values, even better!

As for the rest, you just need to decide what you want for the pool going forward.  Perhaps only a small tweak to doge can help keep miners from jumping ship to that other multi-coin pool after comparing only one or two days of results.  Since you know the value of a doge block in advance, maybe it is appropriate to only go after those with a 2/3 of maximum value or greater, simply in order to help build up your user base at a minimal impact to profit.  Those types of margins are likely to be so thin, that even you might not notice a difference.  Or you might want to keep people away for now if you are not yet prepared with enough capacity, or your database won't yet scale up properly, or countless other factors.  Either way, you're getting a whole lot more right than wrong, and this pool will compete very well.  I was very happy to find it!

I'd prefer staying the course and working for higher potential profits. As this succeeds, the variance issue should resolve itself naturally, since the pool will attract more miners, raising the hashrate, lowering the variance.

Well, that all depends upon whether we are using my special quarter!  poolwaffle, I know the right answer, but it is a pointless endeavor trying to explain
such concepts on a message board.  Everyone likes to believe they have common sense (and misapply it at every opportunity), but it truly is a misnomer as sense is anything but common!

Admittedly though, it did take me longer than I would like to admit to get comfortable with the proper solution to the monty hall problem.  

Understood completely.  The high difficulty is only a temporary measure to keep load on the servers down while we get the new endpoints up.  We'll be making them more forgiving in a few days (miners getting 1024 now will likely be seeing 256ish).

I think a lot of the loss was due to two things.  Its _extremely_ rare that we mine a single coin for that long, since doge (and we're always talking about doge) has variable rewards.  Since it should only take a few blocks to have a low profitability doge block, we would normally switch away, get some other quick profit it, and switch back.  So during those times, while the overall profitability would be slightly lower (unlucky doge time), people would still see their balances going up every few minutes as we grab easier coins in between.

The other reason for the drop in hashrate was one of our current endpoints stopped accepting connections (people mentioning unable to connect), which is what was causing the switcher to stall in the first place (it couldn't connect either).

100% agreed.  I just don't think (and if the community disagrees with me, please let it be known, I'm all for whatever makes this the best pool for you guys) that decreasing potential profits is worth the lowered variance (I guess it depends on how much).  Especially if it can be simply fixed with me not having endpoints crash!

I see your point now it does not matter whether you switch or not chances are still 50 %, you just have to keep flipping the coins long enough and statistically you will get the same result. That really cleared things up. I will switch my miners back to wafflepool thanks

Poolwaffle, the real trouble is that most scrypt miners don't have the understanding or patience to wait for such a long block to be solved.  If their own personal numbers are not constantly moving upwards, then they assume something is wrong and leave the pool to go where their numbers move, even though over time it works out to be the same or better right where they started -- not much unlike those people who incessantly change lanes in heavy traffic.  (And it most certainly can be better on wafflepool as you don't saddle us with a 1024 difficulty and the resulting sky-high reject rates when mining with anything less than the fastest GPUs.  You know who I'm talking about!)

But unfortunately, the pool really does need greater hashpower to go after very difficult doge blocks, _without triggering miner attrition_.  Look at what just happened these last few hours while we were happily stuck mining doge due to a system error -- the pool lost nearly 200MH/s of 850ish capacity .  Perhaps you might consider scaling the profitability calculation for doge and other really high difficulty coins, by some factor of pool hashpower, simply in order to continue building up the pool hashpower, and eventually at some point you will be able to go after any coin you want without losing any miners in the process.

(And yes, I do completely understand that it does not really matter if they leave, that we will all eventually still receive our fair rewards for mining in due time -- but there is also considerable value in being able to reduce variance, even though it does not show up on the long term balance sheet.)

Incorrect.  The chance of finding a block in the next second is exactly the same as the chance of finding a block after mining for 2 hours (difficulty changes disregarded).  Same as flipping a coin. 

Imagine this (reasonably similar) comparison:  You flip a quarter 10 times, you get 10 heads.  One of two things happen next, 1) you switch to a dime, flip it a few times, then switch back to the quarter.  2) You stick with the quarter and flip it again.  Which of these has a higher chance of getting a tails on the last quarter flip?

I will grant you that in the end if your mining a coin with high difficulty it should not matter, whether you mine at a 700 or 7000 MH/s pool but I think for a profit switching pool it is a bit different, because basically you can not really switch to another coin until you have found a block not ? because if you do switch to another more profitable coin you dont get any coins for maybe an hour of mining right ?

On a coin with high difficulty you really have to mine the coin for a longer time period before it evens out when you at a pool with less hash.

This isn't at all correct.  Your hashrate doesn't matter to expected earnings.  Every individual hash you throw at it has a completely static expected value.  Us mining at 700MH/s, and someone else mining at 7000MH/s will have the same (statistically speaking) earnings per MH/s.  The only thing that changes is variance.  We'd expect to go longer between block finds than they would.  And since double the average time is normal (again, statistically speaking), we could go a day without a block, and they might go a few hours.

Even if you were solo mining, the same goes.  If doge is the most profitable, you are best off mining it.  It might take 30 days to find a block, but when you do, the payoff is there.  There are of course extra factors there (difficulty switches, whether or not you want to power a device for 30 days without a reward, etc), but in theory, you're better off mining it.

It wasn't by design....if you read his post.

Frankly I dont think wafflepool hash enough hash to be mining DOGE at the moment, DOGE has the highest network hash at the moment of all the scrybt coins, higher then litecoin even. It takes hours for wafflepool to find a block for DOGE and as long as it has not found a block Wafflepool can not switch, if another coin becomes more profitable because then all the work done on mining DOGE was for nothing.

Hey guys.  Sorry about that, the switcher definitely locked up while I wasn't watching it (dinner).

Essentially we sat mining Doge for the whole time (not the worst time for it to lock up), and while we did find a block, it wasn't a great one.

The reason it locked up is known, and should be fixed now.  Essentially one of our new stratum endpoints locked up while in testing with the switcher (switcher told it to switch, it was like "hang on for a sec", and never came back).  Fixes involve adding much stricter timeouts to the switching of specific endpoints.

Really sorry for the issues, and the delay in getting it fixed.  All part of growing.

You know what would be really nice in the "Time Span / BTC Earned / Avg Hashrate / Daily BTC @ 1MH/s" table, and super easy to implement too?

One new column in between BTC Earned and Avg Hashrate:

BTC Earned (per Day), which is simply BTC Earned (Day X) - BTC Earned (Day X-1)

Or in other words, just the difference in BTC Earned from the line above it.  It makes it really easy for human beings to see how it relates to Avg Hashrate without doing any math.  You could even throw it onto the Stats2 Hashrate graph on a secondary scale and see how they track.

You can keep arguing if you care to, but you will have to do it without me.  I am just going to ignore you starting right now.

I do expect you to try to get the last word in, but I promise you I will not see it.  OK, starting right now!

In that case the stats page is broken.


In that case the switcher is broken too.

We are not mining that block.  But we have indeed been mining doge since then.  And you don't have to look at the stats page to see that.

And how is it that you think it isn't broken?

It claims to be mining a DOGE block of 825,154 and change the whole time. DOGE has a block time of 60 seconds, with rewards that change on every block. How is it we are mining the same 825k coin block for two hours?

It's broken.