That's the problem with all these people. It's not that they can't use Bitcoin, it's that they don't know how to do it properly.
Hint: pay high enough fees and it works. In this particular situation, Bitcoin works as intended.
One may question whether the network is really robust, but the faith of the true bitcoiner will resist any reality attack.
If tomorrow a hacker steals half the coins of the world by exploting some obscure old bug in BitcoinCore that exposed their private keys; or the Chinese miners go on strike and mine only empty blocks, orphaning all other blocks; or the US government blocks all relay nodes except those run by the NSA, that filter out any contributions to Wikileaks or Snowden -- the faithful will still claim that Bitcoin worked as intended. And they woudl be right...
PS. It might not work when Bitcoin actually saturates, and that need to be addressed. But this doesn't have much to do with this stupid attack.
To be pedantic, this was not an "attack" but a "stress test" -- even if a rude and inconsiderate one.
The difference is that a hostile attacker would have tried to block even most of the transactions with higer fees, by raising the fees on its own transactions to keep the legitimate ones perpetually off the next block. And an hostile attacker who decided to do that would probably have a much higher budget, and would prepare his own transactions discreetly months in advance, so that they would be impossible to distinguish from legitimate ones.
I don't know what to think of this test. Since I am not a user, and I see bitcon as a computer science experiment rather than a product ready for general use, I personally think that it was a valid and necesary "stress test" -- a test that one must do on any novel engineering artifact, meant to understand how it behaves if by accident or malice it is used outside its intended operating range (like the stress test that VISA does in August every year -- or the test that caused the Chernobyl disaster
). Such a test must usually be performed on the real thing, since a test mock-up will always lack some feature of it that would break in the test. If possible it should be done "without passengers", but that is impossible for large 24/7 systems like VISA, Bitcoin, or the national power grid. In particular, the goal of the testers of creating a 200 MB backlog on the queues may have been intended to test whether the relay nodes were able to handle queues that big. (And in fact they broke many programs that work on those queues.)
Trying to put myself in place of the typical user, I would think that many who tried to use Bitcoin these days, and were not aware of the need to pay higher than normal fees, must have concluded that the system was broken, and did what any user of free software would do: gave up on it. Note that only a few thousand bitcoin users read the forums or the bitcoin "news" sites (an the latter tend to hide or minimize problems).
It is quite possible that the test had other goals besides a purely technical one. However, it is hard to guess what those intentions could be. The authors are anonymous, and Coinwallet.eu seems to be a phony site create just to give an excuse for the test.
Coinwallet.eu earlier had said that they supported Gavin's original proposal to raise the block size limit to 20 MB. So perhaps they are big-blockians trying to show to the community what a saturated network feels like. They knew that, thanks to the 1 MB limit and ridiculously low minimum fees, they could create a 200 MB backlog (that will take ~5 days to clear) by spending only 5000 euros or so in fees. If the block size limit had been 8 MB, the backlog they could have created with that budget would have been much smaller, and would have been cleared ~15 times faster. However, the test was not very apropriate for that purpose: they did not try to raise their fees in the same rate that normal clients raised their; and they knew that a 200 MB backlog of low-fee transactions would not be as bad as a 20 MB backlog of real ones. Moreover, they did not bother to disguise their transactions, and it seems that some nodes started to reject them.
On the other hand, the testers may have been small-blockians impatient to see and demonstrate the "fee market" that (they claim) will make a saturated network even better than an unsaturated one. Some of the "new devs" invented spiffy tools/toys that they were clearly dying to see in use, but needed a saturated network to work. And some of these devs are notoriously rude and inconsiderate enough to clog the network for days just to see their toys spin and blink.
Or perhaps the "test" was meant to convince the "true bitcoiners" who use the network directly to move to a hosted wallet provider, like Coinbase or Circle -- or perhaps Coinwallet.eu. (Coinbase, for example, seems desperate to get new users: they are now offering 25 $ reward (in BTC) for any user who can convince some friend to register and spend 100 $ with them.) Perhaps even Blockstream is an accomplice of this plan: that would explain why they are so adamantly opposed to any increase in the block size limit, and pretend not to udnerstand what congestion by natural traffic would mean for the usability of Bitcoin...
