The good thing about this thread (one good thing) and about many subjective/objective threads, is that while it's sometimes a bad-tempered argument more than a discussion or conversation, it is still a conversation to some degree. Many threads just die out because because of lack of steam. A bunch of people all going 'isn't so-and-so great' and congratulating each other on their good taste and judgement on buying it can get a bit boring. Being on the objectivist side myself, and not being able to afford $6000 worth of test gear, I sometimes find it a bit difficult to enthuse about my latest creation, because even though I may be well pleased with it, I don't necessarily find a great audible improvement over my last great creation, because (apart from anything else) I generally think that they're all pretty much beyond the threshold of audible distinction. I'm not much given to hearing 'veils lifted', or 'increased texture and detail' or 'greater air and soundstaging'. Over the years I've heard some improvement in the intelligibility of vocals from, say, Mick Jagger or Van Morrison, but this pretty much reached a limit quite some few years ago.
As far as point-to-point or perfboard builds of the O2 are concerned, I think the nwavguy may be being a bit oversensitive about their potential defects. I'm not criticising him in this, he has every right to be defensive about his creation and wish to see it set off only in the best light, but this may lead him to be overly prescriptive about the build.
Some things about any build using his schematic are not going to suffer from the layout. The output current into a given load is pretty much governed by the two driver chips per channel, as is the output impedance, and obviously the peak-to-peak voltage can easily be matched or even exceeded. These are 3 of his main complaints about various marketed designs. The intrinsic noise and distortion due to the components are likewise not subject to any more variation than can be expected by any build on his design of PCB. The noise due to the topology remains the same. The feedback loops are local only, so there is no likelihood of pickup in a global feedback arrangement with a large physical loop area. Instability again is more likely where there are circuit loops with a large area, a neat build is all that is required to avoid these, the chips themselves do not have a huge slew rate and bandwidth, contributing to a reduced tendency to oscillate.
This is not to say that THD + N might not be adversely affected by inductive or static pickup particularly from the power traces or because of a poorly implemented grounding scheme, but if you don't hear any hum, the chances are that there will be no audible difference between a build on his PCB and any other, the standards he has set for THD and noise performance almost certainly exceed the threshold of audibility by some margin. Many people have built cmoys and other headamps with evident satisfaction on perfboard, judging from the enthusiasm with which they defend their builds.
w