We would feel much better being a primary pool

All kidding aside, it's perfectly OK to configure ABCPool as a backup pool. The pattern you describe should not trigger blocking. If you experience any problems though don't hesitate to contact us.

NEWS: Anti-botnet measures kicked in to keep ABCPool healthy
Today for the first time, increased botnet activity on ABCPool has caused our algorithms to block some extremely inefficient users.

Efficiency is the number of shares submitted per requested work. For CPU miners and botnets that number is usually very low. The lower a miners efficiency, the more load it puts on the server. Eventually it begins to affect well-behaving users, who will then see degraded performance.

Since the DDoS attack in January, users below a certain efficiency threshold may be subject to an automatic temporary block. That way they can not cause trouble for well-behaving users. The efficiency threshold is dynamic, based on current conditions.

Blocked users are not able to get work and submit shares. They can still use the website and request payouts.

There may still be glitches in the algorithms that manage blocking. We don't like to block anyone if it is not absolutely necessary. So should you ever discover that you are blocked and don't understand why, please send us a PM and we'll investigate ASAP!

What can you do to prevent being blocked:
* Don't point your botnet at ABCPool
* Don't mine with your CPU. It costs a lot more in electricity than it will every earn you in BTC anyway. Use a GPU.
* Use the latest version of a popular mining client that is known to behave, like cgminer, phoenix, diablominer, guiminer, poclbm, etc.

Happy hashing everyone!

Great suggestion Jjimm, and I've just implemented it for you. Since the info was already on the stats page it wasn't too difficult.

Enjoy!

That might be an option for blocks ABCPool is mining itself. From what I understand it's quite complex to set up right, so not many pools are supporting it. When I read about it (which was months ago) there were also problems for people paying out directly to an exchange. Is that still the case?

Eligius prides itself on being experimental, and p2pool is a different kind of pool altogether. Which mainstream pools put payments in the coinbase transaction?

NEWS: Auto-payout changes to lower the impact of a security breach.
The news that shocked the Bitcoin world earlier today was that several Bitcoin services were compromised after the super-admin password of their provider Linode had leaked. An enormous amount of Bitcoins was stolen. The Bitcoins were stored in wallets on these servers. This allows them to be easily transferred to customers, but also to a potential hacker.

ABCPool was not affected, as we are hosted on Amazon EC2.

Many services (possibly including the services that were hacked today) employ 'cold storage': An off-site wallet that contains the part of the balances that is not directly needed. The cold storage is chosen to be a place that's difficult to reach for a hacker. Cold storage decreases the impact of a hack because only the hot Bitcoins can be stolen. The trick is then to keep the size of the hot wallet as small as possible.

ABCPool has had such a cold storage facility since the security incident we had last year. In light of recent events we felt we needed to review the payment and storage procedures for the pool once more. We've now deployed several changes that make the pool an even less interesting target to hack than before. By making the payment schedule more dynamic, we ensured that our hot wallet is now continuously hovering near 0.

What does this mean for ABC Miners?
* Lower impact in case of a security incident, which means a higher chance of us being able to reimburse you for any losses.
* Payout timing is less predictable. If you're in a rush, request a manual payout.
* Payouts are now processed all day long, not only at 11:00 and 23:00.
* The payout threshold has been replaced by an autopay toggle.
* Payouts are still nicely rounded amounts. Any amount left on your account balance will be part of a future payment.

We will be monitoring payouts closely for a while and play with the knobs to create an optimal balance between security and payment speed.

Let us know what you think or if you encounter any problems.

Happy Hashing everyone!

MC

They're certainly not the same, but there are some relevant shared qualities between the two, like the ones I alluded to above. But then again, I guess every reward method has some shared qualities with every other

In the wake of the Slush pool hack we are evaluating the way payouts are handled at ABCPool. We've already implemented several ways to limit exposure to theft after we were hacked last year. Still we think we can do even more to protect your coins. More info on this will follow later.

Theoretically yes. In practice there are leaks because of orphaned blocks and block withholding. Paying full PPS value is only zero sum if you can eliminate those two factors. Orphaned blocks can be eliminated by retracting payments or paying less than full PPS value; can you point out where it says Ars does any of those?

Ars doesn't pay for orphan blocks? Please provide a source for that statement. I'll provide two that seem to suggest otherwise:

On their front page it says 'You will earn (1 / difficulty) * 50 BTC per share'. A reference in this forum thread that's relevant is:
..which also backs up my explanation: the statistical loss for the invalid block will come out of the buffer, and miners' Lifetime work is not affected. Therefore Ars pays for invalid blocks in essence by taking a loan against the last miners to quit mining there.

The topic was the negative buffer, which i have given a sound explanation for. ABCPool (and your pool too) has some of the same issues that Ars does since PPS and SMPPS are similar, which is why I'm interested in this discussion in the first place.

Ars employs a nominal fee of 0%. The payout is based on 100% of found blocks being valid, but that's unlikely to happen in practice. In practice, some blocks turn out to be orphan blocks. The pool is taking a statistical hit of 50 BTC for every block that turns out to be an orphan, which is around 1.5% of all blocks for most pools.

The effective fee is therefore about -1.5%. Yes, it's an effective bonus for miners! Who pays that bonus? The pool buffer. Total work done at Ars is now about 79850 BTC, which comes down to 1597 blocks. Although I can't find any orphan blocks in the stats at Ars, it's unlikely there haven't been any. You'd expect about 1.5%*1597 = 24 orphan blocks, causing a cumulative buffer drain of about 1200 BTC. The buffer is currently at -1000, not that far from -1200.

Now you may think: Great, the bonus is paid by the pool buffer, so I don't have to care about it. But the pool buffer is just a concept, and eventually real people have got to take the hit and accept the losses of all those orphan blocks. Who will those people be? It will be the last miners to quit Arsbitcoin. Since hashrate will then drop to 0, there's no new BTC flowing in. No income means the payouts for the last -BUFFER_SIZE Bitcoins can never be made.

In a sense, negative-effective-fee-SMPPS is not unlike the game "Musical chairs".

Until the moment the hashrate hits 0, the pool buffer will continue to drain at a rate of 1.5% of the theoretical PPS value of the delivered work, masked by normal variance in the block finding process. It is perfectly possible that due to normal variance the pool gets lucky and the buffer goes into positive territory again. Good and bad luck cancel each other out though, while the 1.5% drain is a constant negative force, creating ever longer payout delays.


Anyway..
For all reward schemes an effective fee can be calculated, and for PPS-derived schemes it differs from the nominal fee. At ABCPool there was some discussion when we started mentioning an effective fee in favor of a nominal fee; some vocal forum members found it to be misleading. Ars is now starting to show just how real the effective fee and its consequences are.
NB: ABCPool is *pure* PPS, there is no concept of a buffer for miners and therefore no payment delays.

Indeed! Even traditional phones gained one: they're now called 'landline telephones'. Analogous to those changes, a pool that proxies shares would then be called a proxy pool.

And how about a pool that proxies part of its shares?

I'm sorry you feel that way. We're trying to balance being upfront with protecting our comparative advantage, just like any other service out there does. Sometimes people accuse us of lying, and it hurts because we have always been honest in our statements.

Being honest does mean we have to keep things a bit vague sometimes to prevent giving any clues to the competition. No other pool than ABCPool has been able to offer such a reliable service for so long with such low risk and such a low fee, and we like to keep it that way.

We've also been honest in our service to our miners. We've always kept any promise we made, and when there were issues we acted in good faith in resolving them quickly and satisfactory.

Now i'd like to get back to the original discussion.

Let me rephrase my answer to the original question so that any perceived evasiveness is gone: 'Yes'. So yes, we mine at other services and send shares to our own bitcoind instances.

Gigavps, I'm eager to hear your response to the arguments of The00Dustin and myself regarding the appropriateness of the 'pool' or 'proxy pool' label for a service like ABCPool.

Good points!

Indeed, when falling back to the plain English meaning of a pool of resources (which people new to Bitcoin will do automatically) the argument gets even more compelling. ABCPool pools miner hashpower together on one end, and pool work suppliers together on the other end.

We have our own bitcoind in the mix too, which comes up with a block now and then.

Nice attempt at a slippery slope argument, too bad it's utter nonsense. Does an ISP split the block reward of the shares that are passed through it? Does it reduce the granularity of the block generation reward? Does it spread it out over time?

The description in the Wiki is actually clear and narrow enough to be a useful guideline.
Before the mobile era, it was widely understood that phones have cords attached to them. Then it turned out that cordless phones can also be used to talk to each other over a distance.

I referenced your service in a neutral way Clipse, no judgement.

I may have been wrong about the nature of your service. I don't know it very well and have only been following the early conversations in your topic. Therefore I retract my statement about Clipse being a proxy pool, and have edited my earlier message to reflect that.

My apologies if I've offended you.

At proxy pools like ABCPool, people still 'pool' their hash power in the common english sense of the word, suggesting that proxy pools are a subtype of mining pools. So ABCPool is a pool that proxies work.

Indeed, ABCPool falls within the definition of 'pooled mining' as featured on the Official Bitcoin Wiki for a long, very long time:
(Quotes from https://en.bitcoin.it/wiki/Pooled_mining)

The page mentions the generation of blocks but does not limit where that generation has to take place. At ABCPool people contribute to the generation of blocks just like at any other pool.

And then there's a more informal argument: I believe it to be ingrained on most miners (and people new to mining) that to decrease variance in earnings, you should join a pool. Decreasing earnings variance is  exactly what ABCPool takes to the extreme by offering PPS rewards, so those people will acknowledge it's a pool. Indeed, in the forums it's common for people to use the term 'pool' when referring to proxy pools like ABCPool, Clipse, Project #2, Multipool, etc.

We are in the business of providing a pool mining service to miners. That's what we promise, that's what we deliver, and that's what our miners are looking for. The fact that part of the work is forwarded is an important detail for some vocal people here on the forum, but not the essence of why people choose to mine with us.

For those that insist on the term 'proxy', at least consider the more informative 'proxy pool' moniker. It recognizes both the pooled nature of pools like ABCPool and the fact that they forward (part of the) hashpower.

EDIT: I may have been wrong about Clipse. To be sure, I'm retracting that statement.

We found the reason of the outage yesterday, and it also explains some earlier problems with stagnation in work supply and share processing.

Some technical details on this: Our code is completely event-based and runs in a single thread. Therefore it's important that at no point there is a blocking function call, or all other tasks must wait for it to complete. It turns out there was a blocking call hidden in one of the libraries we use, that normally returns fast enough not to notice any performance issues, but will under certain conditions block for up to 10 seconds. We've now swapped out the library for a custom version that supports a non-blocking variant of the same functionality.

So what does this change for you?
- Your miner spends even less time idling than before
- Increased pool reliability

UPDATE: Pool is back, but the reason of unavailability is still unknown at the moment.

The pool is having connectivity issues. We're looking into it.

We were indeed posting 'anonymously' using the hollyhunter account.

Using the forum accounts associated with ABCPool for communication with upstream pools could confuse our customers and potentially give away details of our business model. Details that our competitors would be very interested to know. We choose to communicate such practical matters regarding upstream pools through more anonymous forum accounts.

We have contacted Jine privately and hope we'll be able to come to an agreement with him in that way to resolve the issue of the unpaid balances.