I saw this on the news tonight. As a science and statistics guy, I have to say that the methodology of their "experiment" is pretty poor: they used a single trip in each direction, so their sample size is basically nothing. Like you said, it happened to be on a day when GO was delayed by quite a few minutes (which, as a daily GO rider, does happen, but not as often as one might think) and so the TTC came out closer to on-parity than it otherwise would have, given a proper number of trials. Of course, Anne Marie Aikins from Metrolinx was interviewed on that same segment noting the savings of up to ~15 min in each direction that could be expected given an on-time GO, but nobody will believe her after what they just saw.