The way it seems, their solution is working very well given you don't have a stalled network like some other chains being borked with right now... with that said I only see a couple solutions to this kind of attack off the top of my head...
1) Implement merge mining and coordinate with a larger merge mine pool or two to pick you up right away to get a much higher base hash rate making this attack less profitable thus less desirable (but still not impossible and potentially not worth it given other attack vectors scummy players have taken by abusing the merge mining capability)
2) Change some mining parameter around so that ASICS are completely broken on this chain, I would think something even as simple as changing the hash algo to use SHA512 instead of 256 should do it no? There are also the next gen hash algo's that could be looked at and there is a "fall back" to try scrypt or a scrypt merge mine.... asics are not very friendly to change given their nature so the change to break them would be easy. Collude with any exchanges to halt trading and you could even 51% them back by breaking them and mining at a point before they started jacking with the chain and overwrite their blocks with newly fairly generated blocks.
Can the hash algorithm be changed without jacking up the current blockchain though? Would that require a hard fork?
Any change that would be meaningful would most likely require a hard fork, that is a bit more difficult with TRC. TRC is now on a major exchange and if the TRC developers decide to hard fork, there will need to be some communication between the developers and the exchange, to lessen the chance of the exchange dropping TRC.
I have been thinking about this particular problem with ASIC manipulating Alt-coins. I have been trying to think of a tweak of some sort short of the complete scrypt parameter difficulty I have already proposed.
How about this:
Keep difficulty adjustment the same or relativity the same, and run the SHA-256 block hash through scrypt a limited number of times. LTC/NVC use a scrypt 1024:1:1 tuning, roughly this means that the data is run through the scrypt algo 1024x.
What if any of the SHA-256 coins, put a completed SHA 256 hash through a scrypt algo 2x, 10x, or 100x , whatever it takes to cripple the ASICs' efficiency. With the final result, of course, having to meet the difficulty requirement to solve the block. This is subject to the reality of how many times it would take before the coin becomes another LTC like coin. We still want the coin to be GPU friendly, just ASIC hostile.
Downside: mining software developers would have to get involved, or a community patch to a miner to run the SHA-256 hash through the scrypt algo.
Just throwing the idea out there...good or bad