sure, it may start at 0 fee, but for how long? And how long do the xfers take? Seems like normal ACH so a couple days.

It's interesting, but not really a bitcoin-killer.

Hopefully it is better than paypal though.

just got my 3 5830s and they're performing as expected

960/300 on core/mem @ stock voltage and putting out 290Mh/s @ 66C

I thought of an idea, matching peoples bets together so they can bet on the direction of the currency (and earning profit from spread). You can bet over different time frames, IE 1 hour, 1 day, 6 months etc. It will allow people to effectively short the market. The only problem is market manipulation so naturally the maximum bet size would be correlated to the size of the currency somehow.

Thoughts?

Doesn't even seem interesting to me. So it costs zero on the consumer side, so does everything else, right? Charging the merchant is just moving the costs around.

If you ever use PayPal to send money to a friend, why would you ever use PayPal instead of this? For commerce, I don't think it catches on big, but that might not be the point.

clearXchange could be neat if combined with bitcoin. What I mean by that is anyone looking to buy something with bitcoins could send money through clearXchange to another person selling bitcoin, and then pay using the bitcoins they acquired. Merchants or individuals could do the reverse to get dollars. Streamline that process into an app or website and I think you've got a winner.

I suppose the same could be done with paypal, but fees might be higher, and the platform itself is not designed for p2p payments per say. Also like Felix mentioned, just the presence of both paypal and clearXchange in a competitive market would be good for users.

I've done dozens of free transfers to "friends" via PP.

Twice I paid rent to a "friend" for free.

Someone seemed to add almost 3 Th/s instanteously to the network almost doubling it. Either someone either found a cheap way to hash or the government is trying to get involved. I think it is way too risky to get involved in mining now. It is already over $1 a BTC in electricity for me to mine.

not true. see this thread: http://forum.bitcoin.org/index.php?topic=10039.0

after a difficulty change the numbers are all screwy for 24-48 hrs.

I don't know, over 10 blocks in the last hour at the new difficulty. Also the 'other' section has gotten a lot bigger compared to the pools.

I think Bitcoinwatch.com has an accurate hash/sec number and it's way up. Maybe all of those are broken in the hours after a change though, not sure.

apparently it's like this for 24-48 hours after every difficulty bump.

Think about it this way. If you had 3Phash/s of power you could add to the pool, would you do it before a difficulty change or immediately after?

It went from 3.9Th/s to 7Th/s in 1 block. That's not more power, it's just inaccurate reporting.

It also has to do with looking at the recent completed blocks and assuming they are at current difficult instead of previous difficultly. If they actually were at current difficulty and were solved that quickly, I assume the the numbers would all be accurate.

this. it's coming down already.

If I had that much power I'd put in on asap and simply be disappointed if that happened to be right after a big difficulty jump.

I wasn't looking at it at the time of the jump, it sounds like bad reporting. I guess the pie chart is wrong because they use the reported hash speed of each pool and assume the rest is 'other' and total hashing is their wrong number.

I've done it too. But the point is, why bother if you can send direct from account to account without going in the middle? I've had to pay friends through online banking for football pools and other various things, and it's always a pain for them or me to deposit the paper checks that come from it.

I had an interesting thought, and it was "what if you had a way to hash WAY more efficient than everyone else"? To the extreme case, what if you could solve a block within 1 minute almost every time. I figure the smartest play is to wait 6-7 minutes, then start it up. Then wait until 6-7 minutes after every block. That way you don't draw suspicion and no one tries to beat you. You wouldn't solve every block but you could solve a LOT more. Of course, you could just crank away and solve blocks every minute and get the difficulty upped so fast that you'd be the only one capable of solving in the next difficulty, but then it might get ugly. Seems smarter to just keep it going undercover more.

Going regularly timed might give away your strength in a statistical way. Just don't apply all of your power at once imo, let it blend in and look like a general increase in the network.

F that turn it on all at once. suck up as many BTC as you can as quickly as possible. If you aren't sharing your secrets what difference does it make?

you're eliminating your competition that much sooner and can take over most of the mining operations. people will just assume it's big FPGA clusters added to the network and that GPU mining is over.

I'm not exactly sure what hash is being calculated specifically, I think from a quick Google is two rounds of SHA256. I'm not a cryptography expert but by design I don't think there are any short cuts. They would of also been analysed and applied by some really smart people. AFAIK one security feature of hashes is that they are slow to calculate.

I think if there were any performance benefits to be gained it would be by applying the algorithm in assembly if they haven't been already, and these gains would probably be pretty minimal.

Maybe. I guess the worry would be spooking people into thinking that something unfair was going on causing them to stop demanding coins in addition to stopping mining.

You're right that assembly implementations are faster than C. There are implementations in assembly, including ones customized for different processor abilities (SSE, custom instructions, etc). There are implementations for GPU's (OpenCL and CUDA). There's also an FPGA implementation.

Ah ok then, then the only benefit that could be gained then is if you spotted an inefficiency in the algorithm, I'm a fish here but I'm going to be pretty confident in saying that the chance anyone discovers this is going to be extremely minimal.

It was a hypothetical, he isn't suggesting there is a way.

As far as I understand it, there is a way, it's just very expensive. You need to pay TSMC or another high performance semiconductor plant to fabricate a run of custom chips. This is the next step after FPGA iirc.

Wow I hadn't even thought of that, how much faster do people suppose custom chips would be?