GISTIL Sign in

Field notesfrom a measured library

What 3,486 measured videos taught us

We score YouTube videos on how much of the runtime actually delivers. These are the patterns that showed up once there were enough of them to count.

Median padding 40% median start of the useful part 1:00 recalculated 2 Sep 2026

  1. Article1 Sep 20265 min

    Why a summary of a long video reads so well and helps so little

    A summary is a compression, and the things it loses are the things that made the video worth watching. Four failure modes, and the one job summaries are genuinely good at.

  2. Article1 Sep 20265 min

    What Gistil shows you on a YouTube page, and why each part is there

    A verdict, three numbers, and the receipts behind them — shown next to the video before you press play. Here is every part of the panel and what it is claiming.

  3. Article1 Sep 20265 min

    A tutorial from three years ago may not work any more. How to check before you start

    Six minutes into a tutorial the button it names is gone. Here is how to tell which old tutorials are safe, which are not, and why the date is the wrong thing to check first.

  4. Data report1 Sep 20266 min

    The eight-minute pile-up: where runtimes cluster, and what it costs

    Runtimes pile up the moment a video passes eight minutes. The pile is real and easy to see. The penalty everyone assumes comes with it is not.

  5. Data report1 Sep 20266 min

    Sponsor reads take 1.6% of the runtime. Padding takes 40%.

    One in five measured videos carries a sponsor read, and it runs about sixty-seven seconds. The thing that actually spends your time is somewhere else entirely.

  6. Article1 Sep 20266 min

    Where our numbers come from, and where they do not apply

    Nine places where the measurements on this site stop being trustworthy, written down in one page so they do not have to be rediscovered one argument at a time.

  7. Data report1 Sep 20266 min

    Shelf life: how long the material in a video stays true, by subject

    Measured shelf life runs from 84.8 in science to 37.9 in tech news — a 47-point gap. Runtime does not predict it and neither does difficulty. Subject does.

  8. Data report1 Sep 20266 min

    Does a longer video teach you more? We checked, and the answer is yes

    Measured depth rises by 31 points from the shortest videos to the longest, holds inside every subject we can test, and buys less usefulness than you would expect.

  9. Article1 Sep 20264 min

    Do you need an account to use Gistil? Mostly, no

    Reading a measurement we already have takes no account. Asking for a video nobody has measured yet does. Here is exactly where the line sits, and why it is there.

  10. Data report1 Sep 20267 min

    A chapter list tells you where things are, not which one is worth watching

    Section by section across 3,243 videos. The parts a chapter list would call content vary by 34 points inside a single video, and length barely predicts which is which.

  11. Article26 Aug 20265 min

    How to build a watch-later list you actually finish

    The list keeps growing because saving is free and watching is not. One rule, applied at the moment you save, fixes most of it.

  12. Data report26 Aug 20266 min

    The shape of a YouTube video: 3,179 value curves overlaid

    Average measured value climbs fast, holds through the middle two-thirds of a video, then falls sharply near the end. Here is the curve, from 3,179 videos.