Same on Instawallet and Paytunia, and now this: http://blockchain.info/address/1LrPYjto3hsLzWJNstghuwdrQXB96KbrCy

Please not now..

After having used a wallet for a while, my list of unspent outputs grow larger and larger.  Small outputs with thousands of confirmations.  To me it's dust, because I rarely send a payment of less than 1 BTC.  The reason for this is IMHO bad selection of inputs.  The client tries to find the transaction which spends the least inputs, instead of making the transaction with the highest priority.  This costs unnecessary fees as well.

E.g. I recently sent a payment of 3 BTC.  The client found a 3 BTC input with seven confirmations, decided it needed to add a fee and selected an extra input to pay the fee.  The extra input made a change output necessary, and I ended up with a rather low priority transaction with two inputs, two outputs (one very small), and a fee.

The client had many larger and older outputs to choose from, which would have genereated a higher prioirity transaction with one input and two outputs, no output smaller than 1 BTC.  This could have been sent without a fee.

Another possibility is to collect some of the dust (possible inputs < 1 BTC) with thousands of confirmations, adding up to 3.  If the resulting amount is larger, see if it can be reduced to exact by removing inputs.  If not leave a change, or possibly a fee if the extra output would require a fee > change amount.

I think the use-oldest-first method has a few benefits.
1) It is very simple and predictable.
2) Old dust will be collected.
3) In most cases it will generate a higher priority transaction due to higher age of inputs, making the transaction confirm faster.

Comments?

What configuration will work in all cases?  This may be impossible to tell, because it is impossible to prove if bitcoind and all it's components works correctly in all cases.  This follows from The Halting Problem described by Alan Turing.

Testing and avoiding bugs is important, but it is just as important to handle situations like this when they occur, and I think this one was handled very well.  A hard fork with two parallel blockchains going on forever was avoided.  Because bugs will happen.  No amount of testing or algorithm proving can eliminate all bugs.

I have this user on ignore.  I am not alone, as you can see from the piss yellow background of his ignore link.  Nothing even remotely clueful, interesting or constructive comes from his keyboard.

AND, as all critics ignore because they are clueless and it invalidates their criticism, the bug isn't in bitcoin at all!  It is in some versions of BDB on some platforms.  The developers wisely chose a better database for 0.8, but unfortunately the bug manifested itself in a bad way (it could have been worse, e.g. if the bug in BDB made remote code execution or altering old blocks possible) before almost all nodes had upgraded.

AFAIK none of the current developers are involved in BDB development, and BDB was chosen by Satoshi.  Satoshi made a few errors, but I wouldn't call him an idiot.  Shut up or do a better job.

Yes, and it should have the opposite effect.  People should hold their EUR knowing it will buy more BTC (from the USD order book) the next day.  Perhaps even sell for EUR and hope to buy back cheaper because USD is the dominant currency in BTC trading.  Unless they think BTC will be cost at least 0.5% more in USD the next day.  Which means the EUR, GBP and AUD buyers must have very strong belief in price increase.  And I assume the opposite is true for the USD users who are selling, because they are not selling for EUR, GBP or AUD.

Obviously, which is one reason to buy coins.  Deposits of that size triggers identity verification, and if you just use MtGox as a currency exchange between fiat currencies you don't get any anonymity.  Just a very expensive currency exchange.  At least 4% including two trade fees and the known rate change, which is even more than PayPal charges.  People using AUD, GBP and EUR are buying coins, not changing from one fiat currency to another.

This change of sentiment happened after Bitcoin-Central announced their deal with a PSP.  This was even announced on the BBC, and may have increased belief in and knowledge about Bitcoin in Europe, while it seems some Americans have reacted negatively to the news.  Many of my regular customers bought a lot after the news came out (NOK and EUR), and put me temporarily out of funds on the exchanges (doesn't happen often).  AUD users have always been buying more than selling, and it may be that they as well had a positive reaction to the news.   This is just one possible explanation.

That doesn't make sense.  Paying 2.5% + trade fee for currency conversion?  And often when it is already known that the AUD/GBP/EUR price will be much better the next day, making the effective conversion fee more than 3% in the USD -> currency direction?  (MtGox only change their exchange fee once a day.)  Even credit cards have better exchange rates, and the timing is exceptionally bad.

Comparing the price at two exchanges directly doesn't work for several reasons.  BTC-E is always lower as well.  Bitcurex and Intersango are ususally below MtGox EUR price (compared to USD it is more mixed).

Currently the USD is doing much worse than any of the listed currencies.  This may actually be a reason why people choose to keep their own currencies when transferring to MtGox.

I don't think mining is an important factor here.  The volume in those three currencies alone is far higher than the mining rate.  Could be trade balance, if the sellers trade in USD and the people who pay with Bitcoin come from all over the world.  Winter in Europe creates a high demand for alpaca socks, of course. 

I have been watching the trade feed from MtGox for currencies other than USD, and have noticed very different behaviour for traders using USD compared to other currencies.  While the USD traders keep selling, often large amounts in bulk, traders keep buying in other currencies.  GBP, EUR and AUD are especially active.  The prices in GBP, EUR and AUD are almost constantly 2.5% higher, which is where the currency's bid meets USD ask.  (Asks are matched from the USD order book, and exchanged at a 2.5% fee.  This is shown as mixed currency in the feed.)  Large USD sales are often satisfied from bids from the other currencies, and followed by large buys in other currencies.  Recently, when the price hit 13.9 USD, the buying from 13.5 and up was mostly done in non-USD currencies.  Same thing happened when a 10k sale for USD, followed by other large sales in USD, made the price drop below 13 USD.  Large AUD buys pushed the price back up to 13.3 USD.  The last week 18716 BTC was bought in AUD, 10359 BTC bought in GBP and 6767 BTC bought in EUR from the USD order book at a 2.5% fee.  Mostly triggered by bids in those currencies, but also by market sales in USD.  There were no matches in the other direction (where an order from another currency was matched against the bid side of the USD order book).

Is this significant?  Why are the USD users so very pessimistic, while the users of other currencies are so very optimistic?  Is it just dumb users buying BTC in other currencies, while the USD market is speculators selling?  I assume most serious investors would do a currency exchange in their own bank when transferring to get a better rate than the 2.5% MtGox offers.

There is a large number of people from the UK on #bitcoin-otc every day outbidding each other to get hold of coins.  Not very difficult to cash out there.  On MtGox the price in GBP is constantly 2.5% over the price in USD.  I don't know if you have noticed, but it is the GPB (and AUD and EUR) buys that keeps the price falling like a rock at MtGox when people sell thousands of coins for USD.  The overprice at MtGox should cover the fees if you cash out there and transfer the money back to an UK account as a normal international money transfer.

Nobody have to cash out via Euro.  Users can cash out bitcoins any time at no fee, and sell elsewhere for whatever currency they like.

Anyone in Europe (the civilized part) can transfer Euro to you for free or minimal charge.  You can do the same the other way.  SEPA transfers are quite fast.  The regulations say maximum D+1 (banking days), and in my experience same day transfer is quite common if you send early.  D is either the day you issue the transfer, if you do it before an afternoon deadline published by your bank, or the first bank day afterwards.

SEPA bank transfers are not harmed by weird USAnian regulations which forbid you to send money where you want (e.g. to companies with relations to Cuba).  Some bank I tried for one day refused to send money to Intersango because of some issues with an USA based intermediary bank, and I guess you must have a lot of these problems if you are trading all over the world.  (I filed a complaint to my national regulators.  The bank called me the day after with a lot of excuses and sent me a compensation because they hadn't specified the restrictions imposed by the intermediate bank in their terms of service.  Tried the bank once due to free international transfers, but I would rather pay a fee than cope with unreliable service.)

I transfer thousands of EUR to Intersango every week.  How much can you eat?

This is fantastic news!  Welcome as my new Euro bank!

Someone has either messed up EUR and GBP or bid and ask.  Currency mixup and bid/ask mixup happens on Intersango and other exchanges.

Yeah, right, because I'm sure he wants to impersonate you and perhaps even abuse your -19 rating on #bitcoin-otc.  Can I have your ID as well, please?

That's the theory, yes, but it doesn't check out very well when you analyze the orphaned blocks vs the winning blocks.  The advantage to smaller blocks is very small.  There may be one for very small blocks, e.g 1 transaction blocks.

E.g. for the latest orphan, the Bitminter block was 242.8 KB.  The winning block was 250.8 KB, i.e. larger than our.

I have another suggestion: Start making new work as soon as you get a new valid block header built on top of the one we are working on.  The likelyhood of the block being invalid, is low.  This will give us a head start on the next block in front of the other pools.  We may delay LP until the new block is validated.

Are you sure this works?

Last time I did a shallow analysis of this, I found that while there was a connction between block size and the chance of getting orphanded, the larger orphaned block was almost always newer than the winning block.  Which means that the orphan block is larger because it has been accumulating more transactions while the older block was busy propagating through the network.

Smaller blocks means we have to wait longer for normal transactions to get confirmed by the network.  SatoshiDice transactions have a fee, and will win over standard no-fee transactions from normal clients, MtGox, etc.

By reducing the number of no-fee transactions we actually give higher priority to SatoshiDice transactions.  Blocking transactions to or from 1dice addresses with fee < 1 BTC is a better solution, IMHO.

Her er mykje interessant.  Til dømes:
Eit namn har ikkje verkshøgd, og er difor ikkje verna av norsk opphavsrett eller opphavsrett etter Bern-konvensjonen.  Spørsmålet er då kva lover ho tek utgangspunkt når ho gjer dette, og kven som skal handheve desse lovene?  Denne lova gjeld kanskje berre i hennar eigen jurisdiksjon, dvs ein sjølvpålagd restriksjon som ikkje gjeld for nokon andre?

I'll try my question again: How about laptops without Windows?  Windows is an expensive piece malware I don't need, and I certainly don't want to pay for it.