Apart from the claim that it beats competitors, it's a really interesting post, though; light-weight heuristics based on clever connections are interesting in their own right. (So I sort of agree on the tone.)
I'm not the author, not connected with the author, and don't know the author. I found the "paper" and thought "That's neat and elegant."
Yes, we could do with a complete analysis of all existing algorithms and heuristics, finding the strengths and weaknesses, and for each one, finding and characterising cases where they do well and cases where they do badly.
But personally, I like to see papers like this, papers that can be considered to be a "work in progress." I don't want only to read about studies that are exhaustive, complete, and answer all the questions. I think there is value in seeing descriptions of algorithms, along with a simple comparison to known algorithms largely for the purpose of understanding the detail and the potential.
This is an invitation to explore, not a closed and completed "job done." This is an invitation to Google for other heuristics to enhance my understanding of the field. This is an invitation to wonder if my unique range of experience and skills might find a better heuristic.
Full analysis is the work of a Ph.D. and more. Don't expect it in a paper on the web.
While this is cute, I strongly believe that any simple algorithms will compare. Also they seem to hint there is a lack of parallelisable algorithms. Many people have already done work on parallelising TSP by space partitioning.
If so, remember that HN readers tend to be a lot more critical of the article than towards other comments.
Anyway, the article didn't really claim to be the fastest heuristic, just an elegant one with some nice properties. It's O(n log n) - what more do people want? O(n)? O(1)? n log n is almost always fast enough.
http://www.google.com/search?q=comparison+of+heuristics+for+...
It depends on what you actually want, but here are two links that might be the sort of thing you're looking for:
http://www.agent.ai/doc/upload/200302/stut00.pdf
http://www.psych.lancs.ac.uk/people/uploads/TomOrmerod200712...
But let me just say that I found your reply to be unnecessarily nasty in its tone - snippy and snarky. You seem simply to be criticising the authors because they produced a brief paper describing a particular heuristic, and you seem simply to be demanding that they do go further than they have in this paper. Perhaps they already have, perhaps it's work in process.
I find it frustrating when people criticise a link largely because it is what it is and isn't something else. If you want something on comparisons, go find it. Or alternatively, do the work yourself. Then regardless, come and share your results.