Pulse: Call the Trend Before It Peaks

Everyone has an opinion about whether a trend will last. Almost nobody keeps score. Pulse is our attempt to fix the second half of that sentence.

When the batch publishes a full trend report, it also opens one question about that trend — the same question every time, because it is the one we can actually settle: will this query still be among Google↗'s daily trending searches seven days from now? You answer yes or no. That is the whole interaction.

The discipline is in what happens next. The question closes a day before it resolves, so nobody votes with hindsight. On the seventh day the site checks the same Google Trends feed it reads every half hour and records the outcome. Your call is marked right or wrong, and your record — calls made, calls right, current streak — appears quietly under the next question you see.

We were careful about what Pulse is not. It is not a prediction market: there is no money, no odds, nothing to buy or sell. It is not a leaderboard: your record is yours alone. And it is not a poll dressed up as a forecast — we only ask questions the site can resolve itself from a real source, which is why every question shows its resolution source before you answer.

Tallies stay hidden until at least five readers have weighed in, so an early vote never reveals an individual, and the aggregate you eventually see is a real crowd rather than a rumour. One browser gets one call per question.

The reason to do this at all is that reading trends is a skill, and skills improve with feedback. A reader who has called forty trends and been right on twenty-eight has learned something about momentum that no headline can teach. Pulse is the feedback loop.

Sources

• Internal opportunity model · confidence 0.87
• Search Console impressions, trailing 30 days
• First-party trend telemetry, gptautoblog.com