don't forget everyone! BITCOIN CANNOT SCALE! until it does then it still doesn't ... its only been EIGHT MONTHS of arguing over it. #GimpedCoin
Technological progress actually ...ensures that bitcoin not only can scale, but will indeed scale. It's not a matter of if, only a matter of when.
1995 = Pentium 75-100MHz / 8-16MB ram / 1-2GB disks / 14.4kbps - 128kbps connections.
2015 = 4-8core CPUs at 4 GHz / 4-16GB ram / 1-4 TB disks / home connections in the mbps to gbps (for fiber).
Everything has gone upwards like 1000x in 20 years. Processing power, memory, storage, network connections etc. And I'm not counting processing breakthroughs in GPU power or storage speedup like SSDs.
If the trend continues, by 2035 the technology will allow >5000tx/sec, without even improving the software. If the software gets improved (and it constantly is getting scaling improvements), we are talking multiple that - so we'll be seeing VISA-like capabilities way before the 2030's.
Bitcoin is ...doomed to scale, except if some catastrophic failure of our civilization destroys the IT industry. And we won't have to wait for 2035 for scaling, that's just a number. There are incremental steps in hardware and network capabilities all the time. We just want too much, too fast. And that's good because scaling solutions will have to be devised faster. Although one thing will never be overcome: The necessity for fees. It is impossible to have near-zero cost txs AND protect the network for abuse at the same time.
Free txs = attack vector possible for free.
Near free txs = attack vector possible for peanuts.
Hmm...
You know it's really putting your faith in technology evolution and respect of the Moore law...
But experience is showing that Moore law is reaching a limit. The base of Moore principle is that the more time passes, the smaller electronic elements become. But we're reaching a point where you cant get any smaller! It becomes harder and harder and we'll eventually reach a physical limit: you can't go small enough to reach the nanometer. At this scale physics as we know no longer exists...
Thing is, we've seen 1000x in the 1995-2015 step, from things that involve hard disk platter capacity, to network speeds, to processors - so it's not only processors, it's everything. If you told a telco back in '95 that they will be delivering hundreds of mbps over the home copper wire, or a 4G mobile phone, they'd laugh at you. Yet, here we are.
Personally, I expect CPU to become much more efficient in their architecture, even in the same nanometers. X86 chips are generally not very ...optimized and it's the reason why each generation creates big gains over the predecessor with the same clock speeds / similar cache levels etc.
However, with General Purpose GPUs being pushed by the manufacturers, and bitcoin tasks being able to be parallel processed, I think there are great gains that can be made by making bitcoin software that actually takes advantage of the GPU for its processing ability. Mining is probably one of the best known abilities of GPUs, and how instead of waiting for much better CPUs to come out, people went out and created GPU miners that were 10x or more, but there is no reason why their power should not be harnessed for other operations too. I'm thinking specifically of cryptographic operations and real time compression/decompression of small blocks of data (in order to get it to parallelize the operation) - and there could be more.
The biggest gains will be when AI (Artificial Intelligence) starts optimizing both human software and hardware. We may see tremendous boosts or breakthroughs within the next 10 years, and it may be that it is *after* the AI optimizations that we will start seeing a curve of very diminishing returns - as all the huge gains will have been made by the AI.
Cool, so it's time to bump the DoS limit from 2010.
If my cpu or network could endure a flood ping of 100 pings per second in 2010 and now they can endure 500 pings per second, a flood situation will still take me to my limit.
Likewise, you can raise blocksize x10 and an attacker will fill them up if there are cooperating miners that accept cheap spam (and there are).
In my view you just give the space required for legit txs and a bit more for "sudden growth spike" potential, but that's it. Or you give plenty of space right away and impose a high mandatory fee to prevent abuse.