Google I/O came back a couple of weeks ago after skipping a year, and right on cue my timeline filled with people who had decided, within the hour, that everything announced was either the future or a cynical embarrassment. There is rarely much room in between. Keynote season has its own weather, and the forecast is always extremes.
I have watched enough of these now to have lost the ability to be excited by a stage demo, which I think is healthy. The demo is the easy part. Anyone can make a thing work once, on hardware they control, in front of an audience that wants it to work, with an engineer offstage ready to swap the device if it locks up. The demo is marketing. What I have learned to wait for is the bit nobody applauds: whether the thing actually ships, whether it ships to me rather than to a waitlist, and whether it still does the demo's trick six months later when the launch team has moved on.
That is the gap that interests me far more than the keynote itself. The distance between "here is what we announced" and "here is what landed in your hands a year later" is where you actually learn what a company values. Sometimes the unglamorous on-stage line, a quiet platform change or a developer tooling improvement, turns out to matter for years. Sometimes the showstopper demo turns out to be a research project wearing a product's clothes, and you never hear of it again. You cannot tell which is which from the applause in the room.
So I have made my peace with the ritual. The opinions arrive at full volume and full confidence before anyone has touched the thing, which is exactly backwards from how I form opinions about software, which is by living with it until it annoys me or doesn't. None of the takes flying past me this week are wrong, exactly. They are just early. The honest review of any keynote is written about eighteen months after it, by the people who had to use what was promised, and almost nobody bothers to write that one because by then the next keynote is already filling the timeline.
If there is a point here beyond mild grumbling, it is this: treat a keynote as a statement of intent, not a changelog. Intent is worth knowing. It tells you where a platform thinks it is going, which is genuinely useful when you are deciding whether to build on it. Just don't confuse the trailer with the film. The interesting opinions are the ones nobody can have yet.