Bugzilla

I do see positive results with the Early Hints Preload experiment. Ran from Firefox 121. We are now confident in our implementation. The decisionis to move forward with these positive findings and enable Early Hints in release.

```
first_contentful_paint: load time increase by enabling early hints (negative percentages are good for early hints)
0.01: -2.60% = 3.99ms faster
0.05: -4.32% = 10.98ms faster
0.25: -3.69% = 19.66ms faster
0.5: -3.73% = 33.50ms faster
0.75: -2.41% = 36.98ms faster
0.95: -1.04% = 34.62ms faster
0.99: -1.73% = 115.77ms faster
average: -3.16% = 30.19ms
min: 0.01: -2.60% = 3.99ms faster
max: 0.99: -1.73% = 115.77ms faster
page_load_time: load time increase by enabling early hints (negative percentages are good for early hints)
0.01: -3.49% = 8.63ms faster
0.05: -2.43% = 10.59ms faster
0.25: -1.57% = 15.03ms faster
0.5: -1.57% = 27.79ms faster
0.75: -1.25% = 40.07ms faster
0.95: -1.12% = 100.61ms faster
0.99: -1.76% = 452.91ms faster
average: -1.53% = 35.54ms
min: 0.02: -2.66% = 8.29ms faster
max: 0.99: -1.76% = 452.91ms faster
```

I do see positive results with the Early Hints Preload experiment. Ran from Firefox 121. We are now confident in our implementation. The decision is to move forward with these positive findings and enable Early Hints in release.

### first_contentful_paint

Negative percentages are good for Early Hints.

| percentile | relative time difference | absolute time difference |
| ---- | --- | --- |
| 0.01 | -2.60% | 3.99ms faster |
| 0.05 | -4.32% | 10.98ms faster |
| 0.25 | -3.69% | 19.66ms faster |
| 0.5 | -3.73% | 33.50ms faster |
| 0.75 | -2.41% | 36.98ms faster |
| 0.95 | -1.04% | 34.62ms faster |
| 0.99 | -1.73% | 115.77ms faster |
| average | -3.16% | 30.19ms |
| min | 0.01 | -2.60% = 3.99ms faster |
| max | 0.99 | -1.73% = 115.77ms faster |

### page_load_time

Negative percentages are good for Early Hints.

| percentile | load time difference | absolute time difference |
| ---- | --- | --- |
| 0.01 | -3.49% | 8.63ms faster |
| 0.05 | -2.43% | 10.59ms faster |
| 0.25 | -1.57% | 15.03ms faster |
| 0.5 | -1.57% | 27.79ms faster |
| 0.75 | -1.25% | 40.07ms faster |
| 0.95 | -1.12% | 100.61ms faster |
| 0.99 | -1.76% | 452.91ms faster |
| average | -1.53% | 35.54ms |
| min | 0.02 | -2.66% = 8.29ms faster |
| max | 0.99 | -1.76% = 452.91ms faster |

Overall no regressions, just improvements.

I do see positive results with the Early Hints Preload experiment. Ran from Firefox 121. We are now confident in our implementation. The decision is to move forward with these positive findings and enable Early Hints in release.

### first_contentful_paint

Negative percentages are good for Early Hints.

| percentile | relative time difference | absolute time difference |
| ---- | --- | --- |
| 0.01 | -2.60% | 3.99ms faster |
| 0.05 | -4.32% | 10.98ms faster |
| 0.25 | -3.69% | 19.66ms faster |
| 0.5 | -3.73% | 33.50ms faster |
| 0.75 | -2.41% | 36.98ms faster |
| 0.95 | -1.04% | 34.62ms faster |
| 0.99 | -1.73% | 115.77ms faster |
| average | -3.16% | 30.19ms |
| min : 0.01 | -2.60% | 3.99ms faster |
| max : 0.99 | -1.73% | 115.77ms faster |

### page_load_time

Negative percentages are good for Early Hints.

| percentile | load time difference | absolute time difference |
| ---- | --- | --- |
| 0.01 | -3.49% | 8.63ms faster |
| 0.05 | -2.43% | 10.59ms faster |
| 0.25 | -1.57% | 15.03ms faster |
| 0.5 | -1.57% | 27.79ms faster |
| 0.75 | -1.25% | 40.07ms faster |
| 0.95 | -1.12% | 100.61ms faster |
| 0.99 | -1.76% | 452.91ms faster |
| average | -1.53% | 35.54ms |
| min: 0.02 | -2.66% | 8.29ms faster |
| max: 0.99 | -1.76% | 452.91ms faster |

Overall no regressions, just improvements.

I do see positive results with the Early Hints Preload experiment. Ran from Firefox 121. We are now confident in our implementation. The decision is to move forward with these positive findings and enable Early Hints in release.

### first_contentful_paint

Negative percentages are good for Early Hints.

| percentile | relative time difference | absolute time difference |
| ---- | --- | --- |
| 0.01 | -2.60% | 3.99ms faster |
| 0.05 | -4.32% | 10.98ms faster |
| 0.25 | -3.69% | 19.66ms faster |
| 0.5 | -3.73% | 33.50ms faster |
| 0.75 | -2.41% | 36.98ms faster |
| 0.95 | -1.04% | 34.62ms faster |
| 0.99 | -1.73% | 115.77ms faster |
| average | -3.16% | 30.19ms faster |
| min : 0.01 | -2.60% | 3.99ms faster |
| max : 0.99 | -1.73% | 115.77ms faster |

### page_load_time

Negative percentages are good for Early Hints.

| percentile | load time difference | absolute time difference |
| ---- | --- | --- |
| 0.01 | -3.49% | 8.63ms faster |
| 0.05 | -2.43% | 10.59ms faster |
| 0.25 | -1.57% | 15.03ms faster |
| 0.5 | -1.57% | 27.79ms faster |
| 0.75 | -1.25% | 40.07ms faster |
| 0.95 | -1.12% | 100.61ms faster |
| 0.99 | -1.76% | 452.91ms faster |
| average | -1.53% | 35.54ms faster |
| min: 0.02 | -2.66% | 8.29ms faster |
| max: 0.99 | -1.76% | 452.91ms faster |

Overall no regressions, just improvements.