Yes it likely should be sufficient to handle 100x that, assuming the single box was used as dedicated trading engine with other servers used to offload other tasks (API access, webserver, support, charting, etc).

Bitcoin works on the concept of unspent outputs.  If you have two unspent outputs say 1 BTC and 5 BTC and wanted to send 4 BTC you would have to use the 5 BTC and create a 4 + 1 BTC tx w/ the 1 BTC going to a change address.  Likewise if you wanted to spend 5.5 BTC it would be 5+1 as inputs and 5.5 + 0.5 as outputs.  If you however were sending exactly 1, 5, or 6 BTC there would be no change.  The intended output is exactly the size of the input(s).

Only high priority tx are required to pay the min mandatory fee.  The min fee isn't intended as a revenue source, it is intended as an anti-spam/DOS protection system.  High priority tx are not likely to be spam and can be sent without a fee.  Now miners may not include it in the next block but most miners do devote a portion of the block space to no fee high priority txs.  Sending low priority tx without a fee (by hacking the client) can result in tx not being relayed or confirmed for days and sometimes never confirmed.

Name a single person other than you saying that.  I'll wait.  Just a single quote of anyone even a 1 post noob predicting (prior to your post) a $261 exchange rate tomorrow.

Why do you believe a single computer could have mined 2^32 hashes in 10 minutes?  If that were true as soon as two nodes began mining simultaneously difficulty would have increased to 2+.  The fact that it didn't for almost a year combined with the fact that the system took on average 18.5 minutes per block during the difficulty 1 period would also indicate you just made this fact up.  You can't compared the modern miners (which include four years of near continual optimization) to the very "crude" (relatively speaking) miner built into the original client.  It was horribly slow.

The variability in hash power, the long difficulty 1 period, the numerous personal accounts of mining, the historical mailing list records all refute your "facts".  The reality is you have absolutely no fact to back up your lower bound except a grand conspiracy theory.  Stick with math your work there is much better.  This is just sad.

To assist in deepceleron chart.   

To find a block at difficulty 1 requires an average of 2^32 hashes.  So 50 blocks per day = ( 50 * 2^32) / (60 * 60 * 24 * 1000^2) = 2.485 MH/s

 50 blocks per day = ~2.5 MH/s
100 blocks per day = ~5.0 MH/s
144 blocks per day = ~7.2 MH/s
150 blocks per day = ~7.5 MH/s
200 blocks per day = ~9.9 MH/s

Some flaws with the analysis.

1) Difficulty couldn't go below 1 & the "network" has no idea what the hashing power is.  We can only guesstimate the hashing power based on the time between blocks.  If you look at the period of time between blocks in the first year you will notice there is variance.  Variance can either be natural variance from the mining process or changes in hashing power.    It is simply a false statement to claim the hashrate was a constant 7 MH/s.  There is no way to prove that is true and it arguably never was.  The spreadsheet is merely a guesstimate at an aproximate hashrate based on the average time between blocks of the preceding period and is subject to error like any hashrate estimate (i.e the times between blocks was X and since the AVERAGE block requires 4.2 billion hashes the hashrate would need to be ~y to solve block at this rate ON AVERAGE).

2) There is no evidence that Satoshi mined alone the entire time.  Hal Finney reports mining early (in the first 100 or so blocks) and there are other reports of people mining on and off during the early mailing list days.


3) The early client "mined" automatically in the background when running.  For Satoshi to have mined all the blocks for the first year that would mean nobody, not a single person ever downloaded the client which is simply false.

On edit:
4) The first 2016 block interval took 28 days and the next 7 took on average 17 days (with a deviation of only +/- 1 day).  This strongly suggests interest significantly increased after the first week.  One could conclude that the first week was Satoshi alone but I don't think that is likely either.  By May hashing power had fallen significantly  with the 2016 timeframe rising from ~17 days to 28 and then up to a peak of 42.  While there is some variance due to the nature of difficulty to think that Satoshi was the sole miner one would have to think that when the network was at its slowest Satoshi intentionally reduced his own hashing power to roughly half of that in the earlier weeks.   

Please show me two exchanges with a price difference of 30%.

If it existed you wouldn't even need to sell a product.  Just buy x coins on the exchange with lower price and simultaneously sell x coins on the exchange with the higher price.
You raise the price on the lower price exchange, lower the price on the higher price exchange, help reduce the spread and collect an instant 30% profit.

MtGox has proven they intend to be very opaque about the internal problems.  So if you want transparency and they are dedicated to secrecy why do you still use them?

I mean it would be like there is a booth at the carnival which gives punches in the face for $5.  You don't like punches to the face and despite getting a half dozen punches to the face you still plunk down $5 hoping for something better this time.

Maybe it is just the stress of maybe just the Jack Daniels but that made laugh.  It was the second sentence that put it over the top.

This.  If Bitcoin went below a buck the same people complaining about the unfairness of "cheap coins" would run away screaming Bitcoin has failed.  If Bitcoin recovered and in four years was at $500 again they would be again crying that it wasn't fair other people could buy cheap coins.

Hint:  Bitcoins were "cheap" at one time because nobody even knew they were going to be worth a penny in the future.

Yeah it is going to be horrible next week unless MtGox gets their act together.  The sad thing is volume isn't even that high.

Of course MtGox stupidly still allows (as in you can do it right this very second) a user to place for example 1000 x BUY 0.001 BTC @ $100 orders instead of BUY 1 BTC @ $100 orders.  I was convinced in the last lag-freeze this was done intentionally to bring the engine to its knees.  Now on lower volume if someone wants to lag it all to hell they simply need to break their orders even smaller say 10,000 x BUY 0.0001 BTC @ $100.

If someone wants to they could keep MtGox permanently in lag regardless of how much actual volume is occuring.  MtGox does nothing.  No limits on max # of transactions (as opposed to max tx value).  No preventing multiple orders at the same price by the same client, no limiting order generation & cancellation throughput via the API (you can bulk create 10,000 orders then immediately cancel them, then immediately create another 10,000 orders, then immedately cancel them and just keep that loop going forever).

Stop using MtGox?

btc-e, bitfloor, campbx, bitstamp all seem to be working "fine" (some minor website lag and outages at times but nothing compared to MtGox).

It is possible that Bitcoin will not be the dominant crypto currency in the future.  If someone builds something SUBSTANTIALLY superior the free market should adopt that.  Understand however Bitcoin does have a significant network effect.  An "almost as good" or "slightly better in a technical sense that the vast majority could give to craps about" is unlikely sufficient to overcome the network effect.

Ebay was flawed even early on.  No technically superior competitor unseated it though.  This is because for consumers despite the flaws ebay's network effect increased its value more than the flaws reduced it.  As ebay grew the network effect only grew stronger and today it remains the primary auction portal in the world.

Why not use another exchange (even if it is just to sell everything)?  Why do you feel the need to use the worst performing exchange right now ... at a point where it is performing the worst ... and thus add even more load and make it perform even worse?

I honestly think many Bitcoiners have a form of battered wife syndrome "MtGox sucks, so let me use it even more, hoping it will stop sucking more in the future".

Nonsense.  Mining has in the past (and potentially will in the future) operated at negative profitability.  When that happens some miners drop out and the difficulty will adjust.   Bitcoin mining can be profitable at $1,000 per BTC or $1.00 per BTC.   Due to dynamic difficulty it is not like any other industry.

If gold prices fall then marginal miners will become unprofitable, they will stop mining (or scale back) and that WILL REDUCE PRODUCTION.  Lower production = lower available supply = drives prices higher.  As prices rise marginal miners add production.  More production = more available supply = drives prices lower.

With Bitcoin as marginal miners (the least profitable miners) stop mining, difficulty adjusts and the production remains the same.  There is no feedback loop like other industries.

TL/DR price drives difficulty, difficulty never has and never will drive price

If you place a market order then yes it will execute at the best available bid in 45+ minutes which could be 50 or 150.

So you have a couple options:
a) FOR THE LOVE OF SATOSHI .... STOP USING MTGOX.
b) Don't place orders when the lag is more than a few seconds.
c) Use limit orders to ensure your order executes only at the price you want or better.

Over the course of an entire day and reported with timestamps after the closure?

To answer the OP question it seems like MtGox is buffering (delaying) trades on the API.  Ticker info (current bid, ask, last) seems to be updated in realtime but in our own internal analysis we are seeing similar activity.  Polling the trade history API shows no new trades (i.e. pull down trades for last 5 minutes, record last tid, wait 60 seconds, request all trades since last tid, nothing returned = no new trades, try again, nothing returned, try again, nothing returned), then periodically the trade API will "refresh" and show trades for last 2-3 minutes all at once.

Not sure if this is clear but when it "refreshes" the oldest trade since the tid occurred PRIOR to the time that the prior API calls were made and MtGox reported no new trades.

TL/DR I don't think this is a bitcoinity issue, MtGox is either delaying trade activity, not sure if it is intentional, or due to load/lag/DDOS/etc.

Altcoins aren't Bitcoins.  If I launched quadrilla-a-coin which mints 1 quadrillion coins a day forever with no block reward reduction do you think it would massively drop Bitcoin price?  Gold is a store of value, so is silver, and to a lesser extend a handful of other metals.  If people started hoarding uranium it doesn't increase the amount of gold that has been mined.

Fractional reserve "credits" (and you have shown no proof MtGox isn't engaged in it) doesn't increase the supply.  One can also "opt out" by not keeping funds on MtGox (or other hypotehtical fractional reserve bitcoin banks).  Without a central bank (and the ability to print unlimited amounts of funds) any significant fractional reserve bank will eventually fail and when they fail the number of Bitcoin will still be <= 21M.

Transaction fees as a % of miner compensation have increased by a factor of 10x in the past two years.  While still small at ~2% we have a lot of halvings before the reward goes to zero.  It isn't incocievable that fees and transaction volume will slowly rise to fully compensate miners.   Right now if miners received no subsidy the fee would be ~1% of transaction volume.

http://blockchain.info/charts/cost-per-transaction-percent

The blocks (looking at last 2016) are roughly 20% of the 1MB limit so even with no subsidy and only 1MB blocks it seems likely Bitcoin could reach an equilibrium with fees being ~0.2% of transaction volume.  I doubt that is the end game but Bitcoin would still be an impressive value transfer system under those metrics.  Worst case scenario Bitcoin becomes a sort of reserve currency for other lower cost day to day transaction systems.

Yes lost is lost*.  No "lost" coins aren't reintroduced.  There is no way to know if a coin is lost or not.  No, it isn't a problem.  And to answer your next question, no the community isn't going to support a hard fork to remine so called "lost coins" that would be theft.



* In theory it is possible someday ECDSA could be weakened through the discovery of a cryptographic flaw.  This would allow individuals (much like gold treasure hunters) to gain access to lost coins.  Of course we have no idea if or when this might happen.