There would be an unusual number of orphans,  But it would be hard to see because we don't relay orphans, so no single node would necessarily see enough of them to notice.

If bitcoind knows blocks that your application doesn't, then when the next block comes in, track backwards until you find a block that you do know.

Trying to go forward is going to get you into trouble.  You've already discovered orphans.  You also have to deal with reorganizations.

The other way to look at it is that bitcoin is (finally!) the experiment to see if deflationary currency is really bad or not.

In this context, trotting out the well polished argument from authority that "deflation is bad" is extra pointless.

If I were doing that, I would use -blocknotify to trigger my parser, and I'd keep track of the last X block hashes that I've seen.  This would never get stuck, and should be able to handle catching up after being offline for a while.  Just in case though, you could look for the new block's prevHash, and work backwards until you reach a block that you already know.

No matter what, you are going to need some means of dealing with orphans and reorgs, but this way it would be just for your totals.

Interesting approach.  This has some very nice advantages.  For example, we could add a flag to sendtoaddress that makes it return the unsigned tx hex.  Thus we take advantage of the coin selector built into the node, etc.

Dare I ask why you are doing this?

How much does Amazon charge?  Like how much did it cost you to make those graphs, as an example data point.

Chase hopes they found the next step on the stupid patent ladder.

First patent: something possibly useful
Second patent: same thing, but on a computer (80s, early 90s)
Third patent: same thing, but on a network (mid 90s)
Fourth patent: same thing, but on the internet (late 90s, early 00s)
Fifth patent: same thing, but on social media (late 00s, early 10s)
Sixth patent: same thing, but as a cryptocurrency (mid 10s, we are here)
Seventh patent:  same thing, but on (late 10s)
...

Excellent.  A decent launcher is a critical piece for doing useful (aka large scale) simulations.  Do you have any experience with other launchers for big clusters?  As in, can you set it up so that an academic researcher with access to an existing supercomputer array could launch your simulator on their equipment?

Use the one that tells you the opposite of what you want to hear.

If you are bearish and think that failure to cross above the neckline confirms your theory, use the lower line.  If you are bullish and think that successful crossing upwards confirms your theory, use the upper line.

Cool story, bro.

Then go read some old threads.  I'm pretty sure that I've refuted everything you said, when it was said by other people.  My post here was shorthand for "I've totally lost track of how many things in this post are either wrong or undecided".

Oh shit, that's right.  There are actually a whole bunch of coins locked up in scripts that have no possible solution.  Not just OP_RETURN, but also garbage from buggy systems over the years.

"magic internet money"

Yeah, fiatleak is crap.

But...  the last time there were arbitrage opportunities between the bitcoin exchanges, they weren't closed by small bots (heh, like mine) moving coins and dollars around.  They were closed by big traders pushing their positions to multiple exchanges simultaneously.  The current situation looks like that again.  I'd bet a bitcoin that at least one large trader has a system that buys and sells on at least the few bigger exchanges at the same time.  They just aren't using it for proper arbitrage because friction in the legacy banking system makes it difficult to reset positions.

Heh.  Legacy banking system.  I like that phrase...

Quantum computers do not appear to be particularly adept at hashing.

If the network hashrate were steady at say, 1 quadrillion hashes per second*, the difficulty would adjust so that exactly 2 weeks pass per 2016 blocks.  If the hashrate then grows by 10%, blocks will start coming (on average) every 9 minutes, and it will only take 12.6 days to reach 2016 blocks, so the difficulty will go up, by about 10%.

Right now, the network is always growing, so the difficulty is always adjusting up to balance that out.  The more power we add, the higher the difficulty gets.  It is a feedback system fighting to keep the block rate as close to 1 per 10 minutes as possible.

New coin generation comes from blocks.  Because we can't fool the difficulty, the blocks come out on schedule, and thus the new coins come into existence on schedule.  Well, we can fool the difficulty a little.  It doesn't overshoot us, so if we grow really fast on the right day of the cycle, we can bang out a bunch of blocks really quickly before it catches up, which brings future blocks forward in time, more or less permanently, by a bit.

* We are really closer to 5 or 6 quadrillion hashes per second, right now.

Right now fees are sorta related to the size of the transaction, because that is the part that matters, and not the amounts involved.  The amounts in the transaction bear no relationship to the resources consumed by the transaction, so we will probably never switch to that system (but there is nothing stopping individual miners from adopting whatever policy they want).

Take notes on the parts that aren't clear, and if possible, what finally makes it click for you.  The notes and diagrams that etotheipi wrote when he was writing Armory are very useful, and you can help by expanding that.

This is the correct answer.

If you are having problems with it crashing, consider getting a better box.    Or a cron job that checks to see if it is running and restarts it as necessary.