KN,
I think there is a lot of validity and insight to the assertions on the Morrison site regarding nature of recordings and stereo playback.
I would still assert though that recordings are what they are in regards to how miked, mixed etc. Good stereo recordings that apply or approach the two mike configuration sound phenominal on omnis (or pseudo omnis like the Ohms which intentionally attenuate the output towards the walls to make placement easier). This is where there is the biggest gap between omnis and conventional designs, in my opinion. Conventional designs cannot approach omnis, in my opinion, in the ideal scenario, as described by Morrison.
Listen to a good two miked Mercury Living Presence recording on a pair of Ohms, and welcome to a place that few systems have ever gone before.
In the much more common scenario where more complex mikings are used in recordings, omnis will still deliver the image location information better than conventional designs, but the difference is more marginal. Those used to conventional designs will more likely levitate to the imaging inherent in these rather than omnis, because it comes across as more pinpoint, at least in the horizontal dimension (I'd argue about the vertical dimension even in this case, however, and this is where much of the difference is between the omnis and box designs in this case).
I'd take issue with Morrison in describing the more common recording scenario as a "jumble" in that this infers a lack of imaging precision or accuracy, whereas there can be and often is imaging precision in these recordings, however it is based on the whims of the person who did the mix rather than on any inherent natural location of instruments.
Then there are the recordings that are miked and mixed in a more complex manner AND it is done poorly. Traditional box designs will do a better job of masking this due to their inherent imaging limitations even though the garbage is still there. Omnis will let the garbage "shine through", for whatever it may be worth.