Is this correct? If so, this is not acceptable.

CaptainBasculin@lemmy.ml to Firefox@lemmy.ml – 377 points –

Edit: Replies to this thread indicate this is not fully correct as it exists on all browsers; and is likely an ad thing.

37

If the person who tweeted this scrolled down in the hackernews thread, theyโ€™d see this code was misinterpreted. Itโ€™s part of an anti Adblock script that runs 5s after page load. Still shitty, but less insidious

Oh they think that delay in the beginning is malicious .

I noticed a huge drop in bandwidth until I logged out of YouTube. This was a onetimer however.

Don't use the youtube website. On any browser. Use freetube instead, if your situation and use case allows it. No Google bullshit required. Loads fast, no ads, no tracking. Fully self contained.

I really appreciate you saying "if your situation and use case allows it". I'm sick of people "suggesting" these kind of things by just saying "don't do this" "stop doing that" while completely ignoring that not everyone has the same use cases, preferences and possibilities for such.

I wish there was a GTK Piped client. Freetube has that kinda janky feeling of a browser (it's an Electron app, so not surprising) and I don't see why I wouldn't just use Piped in a browser then.

Does freetube show recommendations? Most free clients seem to scrape your subscriptions and lack features like playlists.

Does it proxy requests like piped?

You can proxy through Invidious or use a local API

Whoever posted this is not a programmer. Does no conditional on that code so it would run on every browser on every session so where's the check for Firefox?

Unless they are claiming that it is injected at runtime. But that's easily provable/disprovable with agent spoofing.

In the demo I saw they did an agent spoofing to Chrome and the delay went away, but it didnโ€™t look very extensively tested. As others said, the disappearance on reload could easily be because they thought he was returning to the page and had already seen the ad/been punished for not seeing the ad and so something ad-related disappeared instead.

Iirc the thing is it loads a different js file when it detects chrome which doesn't have the 5s delay. The reasoning is this is part of some anti adblocker code and chrome didn't need the extra logic.

So it's got nothing to do with Firefox it's to do with preventing and blocking so it'll happen on Chrome as well.

Well chrome doesn't need the 5s delay.

Chrome doesn't need a 5 second delay to implement ad blocking, or Chrome doesn't need a five second delay because it's Chrome?

Does an important difference here because one is anti-competitive and the other isn't.

It would not be the first time Google was caught doing this. A couple years ago they were caught breaking apps like google maps if your user agent string wasnt chrome.

But recently I've noticed they can tell regardless of that string. So my guess is that they've hidden fingerprinting code in the chrome browser

I hate how easy is for websites to fingerprint users...

Yep, and it doesn't help that google controls almost all web browsers. They literally give websites your personal information now that cookies are no longer allowed.

So if you're gay and live in Iraq, you better make absolutely sure to only visit your school's website on computer you lend from friends or school computers

This is not correct.

Most of the posts/articles reference following reddit post: https://old.reddit.com/r/firefox/comments/17ywbjj/whenever_i_open_a_youtube_video_in_a_new_tab_its/k9w3ei4/ . It shows the code from your screenshot. However the code does not check the user agent and is not injected server side (I checked by user agent spoofing and using a freshly installed chrome). So it will run on every browser and cannot be used against some specific ones.

There is an answer to the post everyone seems to reference, which goes a bit deeper into what the code could do: https://old.reddit.com/r/firefox/comments/17ywbjj/whenever_i_open_a_youtube_video_in_a_new_tab_its/ka08uqj/

yea there does appear to be delay but it doesnt target firefox users specifically (i think) or atleast isn't obvious to me cuz the code is obfuscated. more likely related to ads (since there is a 5 second delay before you can skip ads). anti-adblock maybe yes.

Laughts in agent spoofing

Browser detection is rarely done through User Agent lookup anymore. Nowadays we determine browser through feature detection.

https://developer.mozilla.org/en-US/docs/Web/HTTP/Browser_detection_using_the_user_agent#avoiding_user_agent_detection

And yet in this case if I change my agent to any non-FF value while using FF, videos load immediately.

I set my agent back to FF, 5 second delay.

Yeah, but some amount of the time is just easier to rely on the user agent. Why bother with the fancy logic when user agent spoofing, adblocking Firefox users is a % of a % of a %?

I tend to agree. I think there's little need as a developer to go that extra mile for accurate browser detection without UA unless it's for fingerprinting. Most feature sets are supported and where it isn't you have a polyfil or whatever shim to make it work. So in the case of fingerprinting you try not to rely fully on anything the user can alter easily.

Doesn't matter what's the best way to do it in reality. UA is easily a possibility for any website. Nothing stopping it

5 more...

I won't spoof the agent. I want Google to know I'll keep using Firefox no matter what. Except for YouTube, I don't use any other service of theirs, too shitty.

The best way to make them back down is to show we won't bow.

The other day, I finally switched to ProtonMail. I still just forward my gmail inbox for now because I'm too lazy to go through all my accounts and change the email all at once. In due time, I will have switched. Fuck Google.

5 more...

uBlock on firefox here, no User Agent spoofing, videos load in ~1-2 seconds as normal still. Wouldn't surprise me if it's used in nicher circumstances though.