First, as I'm sure you know, RTT dominates when it's high enough...and our metrics show it being quite high on these links. For the first few seconds (which is what we care about), you won't get full use of the link's available bandwidth.
-
-
Replying to @slightlylate @fchristant
So RTT matters just as much as channel capacity.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
We have a lot of data sources at Google and, frustratingly, they paint a conflicted picture. I spent most of a week digging into this late last year.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
The biases in the data are also important to tease out: we get disproportionately less Chrome User Experience Report data from users on the slowest links, e.g.
1 reply 0 retweets 1 like -
Replying to @slightlylate @fchristant
Reports like the one you cited require methodological scrutiny. When someone reports observed available capacity that looks suspiciously close to theoretical HSPA+ capacity, something smells fishy.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
In the real world, roughly nobody gets theoretical maximums. That means that if we take the GSMA numbers about connection types at face value (https://www.gsma.com/mobileeconomy/ ), we end up wanting to calibrate expectations conservatively.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
So like I mentioned in the post, we see lots of different things base on the data source and the market.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
Also, data freshness is a challenge here. Some of our internal data sources show median RTTs north of 400ms in the markets growing fastest.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
...but it isn't clear how much to believe any of them. No source has a full picture. The best we can say is that some services with high traffic are able to characterize their usage well in some markets.
1 reply 0 retweets 0 likes -
Replying to @slightlylate @fchristant
...but this has also been sensitive to hysteresis effects: services that drop weight often gain users on slower links.
1 reply 0 retweets 0 likes
That makes me fundamentally suspicious of reports that say "users of our service are on <fast-ish link data here>". Many potential users may simply be unable to access them unless they're built to be lightweight.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
& Web Standards TL; Blink API OWNER
Named PWAs w/
DMs open. Tweets my own; press@google.com for official comms.