1) Here's your periodic reminder that data is time-bound and goes out of date; that much of it was suspect in the first place; that crowdsourced data is even more unreliable; that data is often poorly curated; that you can easily verify all this by moving about in the real world.
2) Data is often uncontextualized. What looks like a traffic jam right now might be a few cars stopped at a red light. What looks like a through road on a generated map might have "no exit" signs all over it. A recommended store in your mapping app may be been closed for months.
3) I've encountered all of these things in the last week, and in no case did I bother to correct them. It takes *unrewarded* time and effort to do that. Let's remember all this as we consider how much of what we do with software of all kinds depends on really solid, timely data.
4) Confirming through testing that everything is OK *now* misses the point. It's essential to question not only the current state of the system, but the assumptions upon which our perception of "OK" is based. What are the consequence when something happens or changes?
5) All this is emphasizes why *variation* in testing is so important, dominating the importance of repetition in testing. Let's not confuse *repeatability* with repetition. The first can be important; the second not so much. Why ask the same question over and over?
