| Commit-Queue | +1 |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
"distance_from_ptr_down_hi", 0.3};I've looked over the (internal only) doc (sorry Yoav!): https://docs.google.com/document/d/1X8_ZnlN5w1NEOTX2gGkZDy5k5xH-MpwW3Jg0Reb-v2Y/edit?tab=t.0
And also spoken to @adit...@google.com who worked on this previously who shared his (also internal-only - sorry again!) Colab: https://colab.corp.google.com/drive/1Wh1Iq3m3Am-l3rZC5Yw5PdMm93jmvinz?resourcekey=0-CXUeV4-JjPFS-Z5cOIrZrA&usp=sharing
Some things to note:
1) The graphs are unevenly balanced. So I don't think it should be -30% to +30%
2) Rerunning the Colab shows the histograms are still available but there are quite coarse and only in 10% buckets. Compared to the doc which shows the same shape of a graph, but with more granularity.
3) Looking at the code (https://source.chromium.org/chromium/chromium/src/+/main:chrome/browser/navigation_predictor/navigation_predictor_metrics_document_data.cc?q=SetDistanceFromLastPointerDown&ss=chromium%2Fchromium%2Fsrc&start=21) it rounds down. So the 0 value is actually covering 0% - 9.99%
Given all that, I think it could be argued that we should set this to `0.1` instead based on that data and update the comment and test appropriately.
Alternatively, we could re-enable the histograms (https://source.chromium.org/chromium/chromium/src/+/main:tools/metrics/histograms/metadata/blink/histograms.xml;l=177;drc=538f2d2b040a8dd2c9be8ec3ae8ef44217c052af) to collect more recent data.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
"distance_from_ptr_down_hi", 0.3};I've looked over the (internal only) doc (sorry Yoav!): https://docs.google.com/document/d/1X8_ZnlN5w1NEOTX2gGkZDy5k5xH-MpwW3Jg0Reb-v2Y/edit?tab=t.0
And also spoken to @adit...@google.com who worked on this previously who shared his (also internal-only - sorry again!) Colab: https://colab.corp.google.com/drive/1Wh1Iq3m3Am-l3rZC5Yw5PdMm93jmvinz?resourcekey=0-CXUeV4-JjPFS-Z5cOIrZrA&usp=sharing
Some things to note:
1) The graphs are unevenly balanced. So I don't think it should be -30% to +30%
2) Rerunning the Colab shows the histograms are still available but there are quite coarse and only in 10% buckets. Compared to the doc which shows the same shape of a graph, but with more granularity.
3) Looking at the code (https://source.chromium.org/chromium/chromium/src/+/main:chrome/browser/navigation_predictor/navigation_predictor_metrics_document_data.cc?q=SetDistanceFromLastPointerDown&ss=chromium%2Fchromium%2Fsrc&start=21) it rounds down. So the 0 value is actually covering 0% - 9.99%Given all that, I think it could be argued that we should set this to `0.1` instead based on that data and update the comment and test appropriately.
Alternatively, we could re-enable the histograms (https://source.chromium.org/chromium/chromium/src/+/main:tools/metrics/histograms/metadata/blink/histograms.xml;l=177;drc=538f2d2b040a8dd2c9be8ec3ae8ef44217c052af) to collect more recent data.
Thanks! I was going off of the public docs, which indicate a distance of 30%, regardless of direction. I guess it'd be good to add that nuance there.
More importantly, at Shopify we're seeing significant gains for moderate on Desktop, but mobile gains are way more modest. So I think it'd be interesting to recollect data on that front (maybe with a split per technology stack), or alternatively to run an OT to experiment with different parameters.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |