3 ms·
I agree complexity should be avoided. But I don't really follow the logic that if every feature can break security, reliability, etc. you should never change a
by matt_oriordan 7y ago
I agree complexity should be avoided. But I don't really follow the logic that if every feature can break security, reliability, etc. you should never change anything or progress. Google is doubling down on these features as you can see at https://www.seroundtable.com/google-pin-live-scores-28261.html https://www.seroundtable.com/google-pin-live-scores-28261.ht.... If they're continuing to invest in adding more realtime features, and penalising sites in search results that don't optimise their websites https://www.sitecenter.com/insights/141-google-introduces-penalty-for-slow-websites.htm https://www.sitecenter.com/insights/141-google-introduces-pe..., then I think they should apply that same logic to their own site IMHO.
- summerlight 7y agoProbably that's why Google is trying to improve network protocols (QUIC, HTTP/3) rather than relying on one-off optimizations described in the article.
- matt_oriordan 7y agoHow does QUIC or HTTP/3 change things? Polling is polling, regardless of whether it's over HTTP/1, HTTP/2 or HTTP/3 (QUIC).
- summerlight 7y agoIt still optimizes overall network efficiency much more than enough to negate 0.001% inefficient cases using polling? I would say this is a right prioritization of optimization engineer headcounts, which is pretty scarce.