nothing new there. any speed advantage, any, no matter how small, is as good as an hour. thats why co location sucks monkey balls.
I am about as worried about someone profiting from creating global/general latency as I was worried about the cookie timing security hole that made the news a few years back. (The amount of time it takes to store a cookie depends upon whether the cookie already exists. Thus a hacked site can sorta snoop what other sites you have been on.) The guy in the article wasn't saying that computer's can track nanoseconds, but that defensive technology (and forensics for that matter) need a fairly substantial anomaly before they want to trigger. One doesn't want an ISP locked-down every time there are 100 zeros in a row transmitted. One needs some level of statistical significance. Most monitoring reports at most once per second. If a directed flood lasts less than 25 ms, it is likely that none of the forensic logs would show anything.
Without arguing the point of whether it is wise, it is none-the-less interesting that there are huge amounts of money being spent on cutting microseconds from the communications path. Not only in co-locating, but in a new undersea cable.
I thought it was a pretty poor article, even if the research behind it might be interesting. The title implies that the idea originated with some nefarious hackers, when in fact the article is based on a "whitepaper" by someone likely trying to sell their services. I doubt it has any practical importance, particularly for anyone but the HFT's themselves (and probably not even for them). They can, at multiple levels, detect discrepancies at arbitrary orders of latency, even if the "current network tools" don't have microsecond resoluntion. It's especially disingenuous that it brings retail traders into the discussion. The typical retail order never even hits an exchange -- almost all of them now go to one of 5 payment-for-order-flow firms. That's a real problem of which we see precious little coverage. http://news.yahoo.com/s/nm/20101217/ts_nm/us_markets_dumb_money