Just read the article, if that didn't make sense -- the author did a much better job of explaining it than I just did.
There is only one counterpoint that really occurs to me, and that is the fact that people are looking for such different things out of games. So, for instance, you might love the pvp in Starcraft, whereas I might not even care at all. I'm interested in how many stars there are in Mario Galaxy 2, because I know I'll collect them all and I want to know how difficult and long that will be (my hope being very difficult and very long); but you might not care one iota.
So I think that fragmentary nature of the readership is one of the drivers behind this "everything and the kitchen sink" type of long-form review. For people who want the details on all the modes and content they might be interested in, all that length is really important. As Rivas says, the individual experience is best described by just a few key moments, but in many games those key moments will vary wildly. Certainly something like Silent Hill 2, Prince of Persia, or Braid is linear enough that probably everybody enjoys (and/or hates) mostly the same parts.
But when it comes to something like an RTS, a 4X, a simulation game, or a many-mode-bearing FPS, that's where things start to get increasingly fragmentary. I might not even play the part of the game you love the most. The only part of Left 4 Dead I've ever played is the co-op, and that in itself was a worthwhile and complete experience for me. Others that want to really make that game a daily event need more, and for those people, there are... some other modes I forget the name of.
This is the conundrum of both reviewers and game creators. Typically the public from-the-developer information about a game is pretty lame and filled with marketing spin. Often it's not even updated once the game is finished and comes out, or at least that used to be the case when I routinely looked at game websites. But the added problem is that many players are naturally skeptical of anything a developer or publisher says about their own titles (I know I sure am), so we want to hear about those things with the slant of an impartial third party. Enter the gaming press, which is therefore left to thus fill the role of both documentarian and reviewer.
All of that stems from the subjective nature of the whole business, I think. With cars or electronics, you can pretty much just relate the hard specifications data, and you can pretty much trust the manufacturer -- because if they lie, it will come out pretty quick and there will be lawsuits. Because all of that sort of thing is pretty objectively measurable. Unlike "fun," "good graphics," "great music," "long," or any of the other adjectives that often get used to describe the various parts of games. All of those terms are either subjective or relative.
My proposed way of reviewing
I think the above is why we have the long-form reviews. In many respects, that sort of documentarian role of the gaming press is super important, and I make frequent use of it. But that tends to make reviews so long that people can't read every last one. Thus we have numbers attached to games, so that players can scan a list of 100 games and choose the 5 they want to read about in more depth.
And, broadly speaking, that makes sense, I think -- we need ways to cull games down so that we can find the ones we are interested in. That's why the concept of "genre" exists, and why those numeric scores came about.
Maybe I shouldn't complain about numeric scores, because my last game was one of the best scored PC titles of last year, but I can't help agreeing with Rivas that the numeric scores are not the best reflection for games. Instead, I think we should have a short-form executive summary (in the style that he admires at Kotaku) for longer-form reviews that provide all that needed extra context. You could even pair that with a very simple pattern of "Avoid, Maybe, Yes" like Nintendo Power tends to do for their downloadable game micro-reviews.
Then if I want to know what games are good, I'd go to a site, and look at the executive summaries for anything in the Maybe or Yes categories for the genres I like. For anything that struck my interest, I'd read the full-length review.
To be honest, the above is more or less what I'm already able to do with the numeric scores (most of which have a breakdown score for each component of the game -- graphics, sound, fun factor, longevity, etc -- along with description of what stands out about each component). But, I think the numeric score implies precision where there is not, and confuses some folks -- certainly causing a lot of distraction both among the press and players. When you have a lot of commentary about a game consisting of "look how many 10s it has!" instead of "look how awesome
But, let's be realistic here; this style of review isn't likely to go anywhere anytime soon, and as far as systems for review go, it's better than many. Try to find out if a mass market novel is any good without reading excerpts from it -- that will drive you to drink.