7 ms·

YouTube page load is 5x slower in Firefox and Edge than in Chrome

107 points by vezycash 8y ago

dagenix 8y ago
It's an interesting topic - but, the linked Reddit thread isn't all that interesting itself. It's most just a collection of unsupported conspiracy theories.
zamadatix 8y ago
The title implies two things: - Google is pushing Shadow DOM before other major browsers agreed to standardize the API - The page load in Firefox/Edge is worse than it would be without Shadow DOM If either or both of these are true then it's pretty shitty on Google's part. If neither are true then it's just a load of FUD. Is anyone more familiar with the situation able to speak to these questions?
- jonathanyc 8y ago
  The tweet is by a well-known Mozilla engineer. In that same tweet he points to an extension that increases YouTube performance by 5x just by reverting to the old design that doesn’t use the Shadow DOM. You can verify for yourself that Shadow DOM v0 is deprecated: https://www.chromestatus.com/feature/4507242028072960 https://www.chromestatus.com/feature/4507242028072960
- Manishearth 8y ago
  For the first one, this is pretty well known : Google "shadow dom v0" v0 is the old stuff chrome shipped before there was consensus, and it's pretty different from v1, which is what the spec settled on. To be clear, "is pushing" isn't exactly correct, it's "pushed". I _think_ everyone is now on the same page wrt v1? (Disclaimer: I do work at Mozilla but I've not been involved in this and this is just what I've read) I'm surprised YouTube is still using v0 since v0 was deprecated a while back.
  - baybal2 8y ago
    Yes, shadow dom + http2
  - ksec 8y ago
    But v1 is not being implemented yet in Edge or Firefox? [1] And then you get these kind of "crap" from twitter [2] It seems we have been talking about Shadow Dom for years, grid, flexbox, web development still isn't anywhere near good enough. [1] https://caniuse.com/#feat=shadowdomv1 https://caniuse.com/#feat=shadowdomv1 [2] https://twitter.com/Shihrer/status/1021891405633925120 https://twitter.com/Shihrer/status/1021891405633925120
    - wener 8y ago
      Nobody see this. Sad
    - dblohm7 8y ago
      v1 is implemented but preffed off in Firefox, IIRC
    - Manishearth 8y ago
      It's implemented, it's preffed off. These things take time to roll out properly. Chrome jumped the gun on v0, Firefox/Edge aren't "dragging their feet".
  - floatboth 8y ago
    > I'm surprised YouTube is still using v0 They're still on Polymer 1.x. Upgrading takes time I guess.
kristianp 8y ago
The tweet that it refers to: https://twitter.com/cpeterso/status/1021626510296285185 https://twitter.com/cpeterso/status/1021626510296285185
jcstryker 8y ago
If true, then it is yet another example of anti-competitive behavior from Google. Reminds me of Firefox support in Hangouts.
- thermodynthrway 8y ago
  The worst I've seen is that Google serves an ancient version of their homepage to Firefox mobile for "reasons". With an extension to spoof useragent to Chrome, you get the regular page and it works fine
  - leppr 8y ago
    Used to be the case with Google Maps as well, Chrome user agent would get you served with a faster version. The worst I've seen though isn't by Google themselves. It's user-agent whitelisting on other web sites, that think it's okay to prevent users from using the browser they want because they couldn't spend time testing in the other browsers. Sometimes they even exclude Chromium. For that reason I have an user agent spoofing extension on each of my browsers.
    - thermodynthrway 8y ago
      IMO, Google put a new stage on embrace extend extinguish I call "neglect". Make everything open source and nice, then continue to update G specific functionality by simply not maintaining good support for anything else. Google probably has the best engineers and greatest minds since IBM in the 60's, possibly ever. But great engineers are attracted to shiny objects, and if all of those are Google objects there's suddenly no reason to use anything else
squirrelicus 8y ago
I'm not super quick to think they're doing something malicious here. I've experienced this exact thing at two of my jobs this decade, one of which I currently work at. Firefox and especially Edge are often just slower for some DOM manipulation use cases and style recalculations. Although Pendo has been identified as the primary cause for Edge slowness :P
- jonathanyc 8y ago
  No, the tweet points out how this is specifically because Google uses a deprecated, older version of a standard that only Chrome supports. It’s absolutely not a question of Firefox or Edge being “just slower,” it is a question of using a feature that is literally only implemented in Chrome and that other browsers have agreed not to implement.
  - squirrelicus 8y ago
    Thanks for the correction. Clearly I should not have skimmed so lightly.
  - detaro 8y ago
    And the current version of the standard also isn't implemented in Firefox yet, so using that would change nothing?
    - dblohm7 8y ago
      It is implemented behind a preferred ATM, IIRC
- andrewmcwatters 8y ago
  I think in any stride in life you can easily make decisions that aren't well thought out that are malicious without intent.
gaius 8y ago
Time for antitrust to break Google up. At the very least force the divestiture of both Chrome and YouTube. And obviously given the case in Europe, Android as well.
HNNewer 8y ago
Sure, on Firefox adblocking works pretty much solid, removing all kind of ads thanks to uBlock origin. On Chrome, I get sometimes still ads across videos, plus Chrome is Google's Browser: obviously they test YouTube extensively with it and they just make it "work" for other vendors.
- ferongr 8y ago
  uBlock Origin shares the same code in Firefox and Chrome. Any differences you see have nothing to do with the browser.
  - krageon 8y ago
    They do, because there is an exploit for chrome that makes 3rd party look like first party. https://github.com/gorhill/uBO-Extra https://github.com/gorhill/uBO-Extra You will only get this behaviour on Chrome, Firefox is not vulnerable.
zaarn 8y ago
And now I know why Youtube always takes several seconds to begin working after page load. Usually it would sit there, showing me a thumbnail or even daring to play an add while basically every single UI element was a grey blob slowly pulsing while the javascript in the background was busy shitting data into a polyfill. And with that addon? Basically instant page load. Why in all that is holy did anyone think this was a good idea and why did noone bother to test this on Firefox or if they did why did the engage in such shitty behaviour?
- M2Ys4U 8y ago
  >Why in all that is holy did anyone think this was a good idea and why did noone bother to test this on Firefox Because Google only cares about making Chrome look good. Take a look at pretty much _every_ Google property and at some point or another they've fucked over other browsers in order to make Chrome more attractive for end-users.
  - zaarn 8y ago
    I don't think that most google employees do this in a malicious or active manner. But I do think management encourages such outcomes. I'm basically no longer going to support Chrome on my projects, if it works, yay, if it doesn't, get a real browser.
    - tmd83 8y ago
      Honestly what should one consider it then, gross incompetence. I work in a fairly small company, we have 100s of unique forms in our application and we try to make it work and work fast in 3 browsers and sometime we do find/gets reported on performance problem at specific things and we fix those. How many unique pages or functionality the youtube site has. Am I missing something here? The scaling problem of youtube backend is amazing, complicated and unique. What is the unique uber complicated functionality does the youtube page itself has that they can't test their main page loading up 5 times slower in 2 of the 3 browsers? Do they even open it another browser?
megaman22 8y ago
I would hope that Googlers can make their sites work fastest on the browser that they own and control the internals of. If I were building a browser, I'd be seriously tempted to do a lot of things for speed that don't seem to be on the table right now. For instance, I'd aggressively bundle common JS or CSS dependencies and short-circuit loading them over the wire. E.g. a site wants jquery 1.x, or Bootstrap 3? I can load that faster from even spinning rust than over fiber, and a few MB of JS or CSS is nothing to store.
- fps_doug 8y ago
  > If I were building a browser, I'd be seriously tempted to do a lot of things for speed that don't seem to be on the table right now. Yes that's all fine, but we're talking about YouTube here as the offender not Chrome. It's fine to make your browser work faster, that's not the same as making your website work slower on other browsers. > I would hope that Googlers can make their sites work fastest on the browser that they own and control the internals of. I would hope they'd do that by just optimizing for Chrome, not by deciding to base the whole website on an API that's deprecated and has been superseded and is ONLY still supported by Chrome.
- hadrien01 8y ago
  > For instance, I'd aggressively bundle common JS or CSS https://addons.mozilla.org/en-US/firefox/addon/decentraleyes/ https://addons.mozilla.org/en-US/firefox/addon/decentraleyes...
wruza 8y ago
It I remember when chrome only appeared, they deployed try-chrome banners, specifically seen on /. that froze both opera and firefox for a couple of seconds, but worked fine in chrome itself. No one believed me on local forums back then.
- hadrien01 8y ago
  I think you messed up your example url.
  - watermelon0 8y ago
    I assume he meant https://slashdot.org/ https://slashdot.org/
mabynogy 8y ago
I confirm that here too with debian and firefox 52.9. YT is damn slow and HookTube is faster even with the embedded YT player.
skybrian 8y ago
So apparently a misguided dependency on an obsolete version of a web framework that uses a dodgy polyfill on some browsers is now proof of a dark conspiracy since browsers won't have the same performance. This a good discovery and certainly an unfortunate result, but do people not know how web development works? Mistakes like this do happen (though usually with less severe consequences).
- ledriveby 8y ago
  HN hivemind: the only cure for a Google bug is... more regulation!
- someonewhocar3s 8y ago
  I'm sorry, but Google implementing unspecc'd features and then using these features such that it causes a 5x slowdown in (actually) faster browsers... I guess you were happy with IE back in the day too? You just find such anti-competitive behaviours acceptable? Or are you just wealthy enough to buy a 5x bigger PC (consume 5x the power...) to just watch youtube with the same experience as a superior webbrowser would offer? I think it makes a lot of sense to ask for a standard for companies to build their products against, such that they become naturally unable to abuse the ties between their distinct products, to avoid vender lock-in following a diverged software stack. Standards are n i c e

107 points by vezycash 8y ago

It's an interesting topic - but, the linked Reddit thread isn't all that interesting itself. It's most just a collection of unsupported conspiracy theories.

zamadatix 8y ago

The title implies two things: - Google is pushing Shadow DOM before other major browsers agreed to standardize the API - The page load in Firefox/Edge is worse than it would be without Shadow DOM If either or both of these are true then it's pretty shitty on Google's part. If neither are true then it's just a load of FUD. Is anyone more familiar with the situation able to speak to these questions?

jonathanyc 8y ago
The tweet is by a well-known Mozilla engineer. In that same tweet he points to an extension that increases YouTube performance by 5x just by reverting to the old design that doesn’t use the Shadow DOM. You can verify for yourself that Shadow DOM v0 is deprecated: https://www.chromestatus.com/feature/4507242028072960 https://www.chromestatus.com/feature/4507242028072960
Manishearth 8y ago
For the first one, this is pretty well known : Google "shadow dom v0" v0 is the old stuff chrome shipped before there was consensus, and it's pretty different from v1, which is what the spec settled on. To be clear, "is pushing" isn't exactly correct, it's "pushed". I _think_ everyone is now on the same page wrt v1? (Disclaimer: I do work at Mozilla but I've not been involved in this and this is just what I've read) I'm surprised YouTube is still using v0 since v0 was deprecated a while back.
- baybal2 8y ago
  Yes, shadow dom + http2
- ksec 8y ago
  But v1 is not being implemented yet in Edge or Firefox? [1] And then you get these kind of "crap" from twitter [2] It seems we have been talking about Shadow Dom for years, grid, flexbox, web development still isn't anywhere near good enough. [1] https://caniuse.com/#feat=shadowdomv1 https://caniuse.com/#feat=shadowdomv1 [2] https://twitter.com/Shihrer/status/1021891405633925120 https://twitter.com/Shihrer/status/1021891405633925120
  - wener 8y ago
    Nobody see this. Sad
  - dblohm7 8y ago
    v1 is implemented but preffed off in Firefox, IIRC
  - Manishearth 8y ago
    It's implemented, it's preffed off. These things take time to roll out properly. Chrome jumped the gun on v0, Firefox/Edge aren't "dragging their feet".
- floatboth 8y ago
  > I'm surprised YouTube is still using v0 They're still on Polymer 1.x. Upgrading takes time I guess.

kristianp 8y ago

The tweet that it refers to: https://twitter.com/cpeterso/status/1021626510296285185 https://twitter.com/cpeterso/status/1021626510296285185

jcstryker 8y ago

If true, then it is yet another example of anti-competitive behavior from Google. Reminds me of Firefox support in Hangouts.

thermodynthrway 8y ago
The worst I've seen is that Google serves an ancient version of their homepage to Firefox mobile for "reasons". With an extension to spoof useragent to Chrome, you get the regular page and it works fine
- leppr 8y ago
  Used to be the case with Google Maps as well, Chrome user agent would get you served with a faster version. The worst I've seen though isn't by Google themselves. It's user-agent whitelisting on other web sites, that think it's okay to prevent users from using the browser they want because they couldn't spend time testing in the other browsers. Sometimes they even exclude Chromium. For that reason I have an user agent spoofing extension on each of my browsers.
  - thermodynthrway 8y ago
    IMO, Google put a new stage on embrace extend extinguish I call "neglect". Make everything open source and nice, then continue to update G specific functionality by simply not maintaining good support for anything else. Google probably has the best engineers and greatest minds since IBM in the 60's, possibly ever. But great engineers are attracted to shiny objects, and if all of those are Google objects there's suddenly no reason to use anything else

squirrelicus 8y ago

I'm not super quick to think they're doing something malicious here. I've experienced this exact thing at two of my jobs this decade, one of which I currently work at. Firefox and especially Edge are often just slower for some DOM manipulation use cases and style recalculations. Although Pendo has been identified as the primary cause for Edge slowness :P

jonathanyc 8y ago
No, the tweet points out how this is specifically because Google uses a deprecated, older version of a standard that only Chrome supports. It’s absolutely not a question of Firefox or Edge being “just slower,” it is a question of using a feature that is literally only implemented in Chrome and that other browsers have agreed not to implement.
- squirrelicus 8y ago
  Thanks for the correction. Clearly I should not have skimmed so lightly.
- detaro 8y ago
  And the current version of the standard also isn't implemented in Firefox yet, so using that would change nothing?
  - dblohm7 8y ago
    It is implemented behind a preferred ATM, IIRC
andrewmcwatters 8y ago
I think in any stride in life you can easily make decisions that aren't well thought out that are malicious without intent.

gaius 8y ago

Time for antitrust to break Google up. At the very least force the divestiture of both Chrome and YouTube. And obviously given the case in Europe, Android as well.

HNNewer 8y ago

Sure, on Firefox adblocking works pretty much solid, removing all kind of ads thanks to uBlock origin. On Chrome, I get sometimes still ads across videos, plus Chrome is Google's Browser: obviously they test YouTube extensively with it and they just make it "work" for other vendors.

ferongr 8y ago
uBlock Origin shares the same code in Firefox and Chrome. Any differences you see have nothing to do with the browser.
- krageon 8y ago
  They do, because there is an exploit for chrome that makes 3rd party look like first party. https://github.com/gorhill/uBO-Extra https://github.com/gorhill/uBO-Extra You will only get this behaviour on Chrome, Firefox is not vulnerable.

zaarn 8y ago

And now I know why Youtube always takes several seconds to begin working after page load. Usually it would sit there, showing me a thumbnail or even daring to play an add while basically every single UI element was a grey blob slowly pulsing while the javascript in the background was busy shitting data into a polyfill. And with that addon? Basically instant page load. Why in all that is holy did anyone think this was a good idea and why did noone bother to test this on Firefox or if they did why did the engage in such shitty behaviour?

M2Ys4U 8y ago
>Why in all that is holy did anyone think this was a good idea and why did noone bother to test this on Firefox Because Google only cares about making Chrome look good. Take a look at pretty much _every_ Google property and at some point or another they've fucked over other browsers in order to make Chrome more attractive for end-users.
- zaarn 8y ago
  I don't think that most google employees do this in a malicious or active manner. But I do think management encourages such outcomes. I'm basically no longer going to support Chrome on my projects, if it works, yay, if it doesn't, get a real browser.
  - tmd83 8y ago
    Honestly what should one consider it then, gross incompetence. I work in a fairly small company, we have 100s of unique forms in our application and we try to make it work and work fast in 3 browsers and sometime we do find/gets reported on performance problem at specific things and we fix those. How many unique pages or functionality the youtube site has. Am I missing something here? The scaling problem of youtube backend is amazing, complicated and unique. What is the unique uber complicated functionality does the youtube page itself has that they can't test their main page loading up 5 times slower in 2 of the 3 browsers? Do they even open it another browser?

megaman22 8y ago

I would hope that Googlers can make their sites work fastest on the browser that they own and control the internals of. If I were building a browser, I'd be seriously tempted to do a lot of things for speed that don't seem to be on the table right now. For instance, I'd aggressively bundle common JS or CSS dependencies and short-circuit loading them over the wire. E.g. a site wants jquery 1.x, or Bootstrap 3? I can load that faster from even spinning rust than over fiber, and a few MB of JS or CSS is nothing to store.

fps_doug 8y ago
> If I were building a browser, I'd be seriously tempted to do a lot of things for speed that don't seem to be on the table right now. Yes that's all fine, but we're talking about YouTube here as the offender not Chrome. It's fine to make your browser work faster, that's not the same as making your website work slower on other browsers. > I would hope that Googlers can make their sites work fastest on the browser that they own and control the internals of. I would hope they'd do that by just optimizing for Chrome, not by deciding to base the whole website on an API that's deprecated and has been superseded and is ONLY still supported by Chrome.
hadrien01 8y ago
> For instance, I'd aggressively bundle common JS or CSS https://addons.mozilla.org/en-US/firefox/addon/decentraleyes/ https://addons.mozilla.org/en-US/firefox/addon/decentraleyes...

wruza 8y ago

It I remember when chrome only appeared, they deployed try-chrome banners, specifically seen on /. that froze both opera and firefox for a couple of seconds, but worked fine in chrome itself. No one believed me on local forums back then.

hadrien01 8y ago
I think you messed up your example url.
- watermelon0 8y ago
  I assume he meant https://slashdot.org/ https://slashdot.org/

mabynogy 8y ago

I confirm that here too with debian and firefox 52.9. YT is damn slow and HookTube is faster even with the embedded YT player.

skybrian 8y ago

So apparently a misguided dependency on an obsolete version of a web framework that uses a dodgy polyfill on some browsers is now proof of a dark conspiracy since browsers won't have the same performance. This a good discovery and certainly an unfortunate result, but do people not know how web development works? Mistakes like this do happen (though usually with less severe consequences).

ledriveby 8y ago
HN hivemind: the only cure for a Google bug is... more regulation!
someonewhocar3s 8y ago
I'm sorry, but Google implementing unspecc'd features and then using these features such that it causes a 5x slowdown in (actually) faster browsers... I guess you were happy with IE back in the day too? You just find such anti-competitive behaviours acceptable? Or are you just wealthy enough to buy a 5x bigger PC (consume 5x the power...) to just watch youtube with the same experience as a superior webbrowser would offer? I think it makes a lot of sense to ask for a standard for companies to build their products against, such that they become naturally unable to abuse the ties between their distinct products, to avoid vender lock-in following a diverged software stack. Standards are n i c e