A Conspiracy to Kill Ie6 (2019)
Key topics
The story of how YouTube engineers conspired to kill IE6 by displaying a warning message to users, sparking a discussion on the impact of this decision on the web and the trade-offs between supporting older browsers and driving innovation.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
36m
Peak period
109
0-12h
Avg / period
19.9
Based on 139 loaded comments
Key moments
- 01Story posted
Oct 16, 2025 at 2:22 PM EDT
3 months ago
Step 01 - 02First comment
Oct 16, 2025 at 2:58 PM EDT
36m after posting
Step 02 - 03Peak activity
109 comments in 0-12h
Hottest window of the conversation
Step 03 - 04Latest activity
Oct 21, 2025 at 2:10 AM EDT
3 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
> Frustrated, one of the lawyers asked “Why did you have to put Chrome first?” Confused, I explained that we did not give any priority to Chrome. Our boss, in on the conspiracy with us, had thoughtfully recommended that we randomize the order of the browsers listed and then cookie the random seed for each visitor so that the UI would not jump around between pages, which we had done. As luck would have it, these two lawyers still used IE6 to access certain legacy systems and had both ended up with random seeds that placed Chrome in the first position. Their fear was that by showing preferential treatment to Chrome, we might prick the ears of European regulators already on the lookout for any anti-competitive behavior.
Wow those lawyers must've left the place many years ago huh!
I remember that the former GM of the Internet Explorer 5 and 6 team transferred to my org about a year after I joined. In his intro email, he included a sheepish apology for IE6, which I printed and kept on my office wall for the rest of my time at Bing, it was a prized possession. Man that browser caused so many nightmares.
(to clarify, the GM was a good and smart guy, the apology was a little tongue-in-cheek since IE6 was arguably the best browser upon its release - the problem was Microsoft effectively abandoned it and let it languish and stagnate for years while the web moved on without it, which turned it and the IE org into well-deserved pariahs)
For comparison, Internet Explorer 6 came 2.5 years after 5 and so did 8 after 7.
The lesson of IE6 is that people cannot be trusted to handle updating themselves.
The only good justification I can think of to update totally working software that I am happy with, is for mitigating security vulnerabilities. And even then, the choice should be on the user.
Why evolve software at all?
It often didn't, but the user was not always in control. Many business and educational environments held back and their users were in locked down machines (for good reasons) so could not upgrade if they wanted to.
This had a secondary effect: parents in households where the kids weren't in control of the tech were wary to upgrade in case it made them incompatible with work or things the kids needed for school.
> Maybe if IE6 was so terrible, Microsoft shouldn't have released it in the first place.
As has been mentioned a few times, upon release IE6 was the best browser commonly available by a number of metrics. Netscape was properly stagnating around then, Firefox was not yet a thing (even under is earlier names), chrome was even further off, and other alternatives only captured a niche market. A lot changed between then and 2009 but IE6 didn't.
> this assumption that we should always have to be on the update treadmill
This wasn't the enshitification treadmill that we experience today. Newer browsers at the time were offering key benefits for performance and security as well as significant useful features for designers that had to be inefficiently polyfilled or rejected if you needed to support older UAs.
> mitigating security vulnerabilities. And even then, the choice should be on the user.
No. As much as I disparage Windows for being the OS that can't be trusted not to randomly reboot if you leave it unattended for 12 hours, security updates are everybody's problem if you get infested with something that goes on to affect the wider network.
The rogues take responsibility, think carefully, act carefully.
I do occasionally think Safari is the new IE though -- not in terms of terribleness but just in terms of holding back the web by being the slowest to implement big new features.
You mean Chrome-only non-standards that Mozilla usually opposes, too
I'm sure there are many such examples.
Firefox implemented them in 2014, Edge in 2020, and Safari re-introduced them in 2022.
The was a proposal to remove the spec entirely in 2015: https://github.com/whatwg/html/issues/315
I don't think I've seen a feature that was shipped in s browser for years, then removed, and then shipped again.
I'd love to know why. They only explanations I've seen seem to be this: https://x.com/xeenon/status/652573047623323648 "The implementation of Shared Web Workers was imposing undesirable constraints on the engine. It never gained any adoption." and this: https://bugs.webkit.org/show_bug.cgi?id=149850#c5 "This feature was originally removed temporarily during multiprocess bring-up, and actual usage on the web has been pretty low. We're willing to reconsider if there is significant demand."
Since 2014, three years after desktop, I'll grant you that (though it's not forever)
> WebGPU still is only partially supported on desktop.
And it's only partially supported in Firefox. Perhaps something to do with actually fleshing out the APIs and not rushing with implementations like Chrome?
Edit: Firefox enabled WebGPU only on Windows this year.
Safari enabled WebGPU on all devices this year with *OS 26.
> Memory64 isn't available yet at all.
You mean the one released in Chrome this past February, and in Firefox this past January? Less than a year old?
And used on a whopping 0.0001% pages? https://chromestatus.com/metrics/webfeature/timeline/popular...
And honestly I wouldn't care except they don't let you USE ANY OTHER ENGINE in mobile. If I could just skip Safari I would, but even Chrome on iOS is basically Safari.
Often for s good reason. Once something is in a browser, there's signs no change in hell to fix it or remove it.
Chrome doesn't care. They just ship whatever even if the other vendors are actively against it for many reasons.
https://www.justice.gov/archives/opa/media/1344546/dl
People try to equate it to Safari now but that's just not comparable. Safari will render something badly or not support a CSS decorator that you'd really like to use, but it will rarely crash, go into an infinite URL-fetching loop, or arbitrarily fail to recognize random HTML tags.
There was a sweet spot between roughly 2007 and sometime in the mid 2010s when web developers coded to standards instead of just the dominant browser, and where there was browser diversity: Firefox, Safari, Opera, Chrome, and IE 7+. It was a good time for the Web.
Chrome then became dominant, and unfortunately now we're in a "Best viewed in Chrome" era, and we're back in an era where some developers only code for the dominant browser.
Chrome when it came out was much faster than Firefox. It was lighter and worked better.
Also macs had moved to Intel chips a few years before and were actually pretty decent so a lot of people were moving to them.
Also of this chipped away at XP and the few people running XP machines were diehard xp fans or corps that were dragging their heels upgrading.
It also had an outsized impact on the web because it was a popular with developers for doing web development.
In 2010 Firefox 3.5 had a estimated global market share of 15-20%. BTW nobody thought any of the stats were accurate at the time as they were easily skewed. Some counters would report 30%, but it was an estimate and not a reliable one.
On the sites I was building, which were mostly travel sites, e-commerce and later gambling. Firefox was maybe 5-10%. I am also no in the US. I just didn't see anything like what was reported in site stats for the things I was working on.
Later on it was IE and Chrome and Firefox was still at maybe 10%. I really cared about compatibility and web standards at the time and made every effort to make sure that the sites work. So I knew it wasn't a "the site doesn't work in this browser".
> It also had an outsized impact on the web because it was a popular with developers for doing web development.
So people that used the net heavily used Firefox and people that didn't tended to use IE. So on some sites Firefox usage would be far higher than it would otherwise be.
e.g. People using IE might only use the internet for online shopping, checking mails, so visiting an online shop, booking a flight etc. Whereas many Firefox users would be using Social media, Blogs, Forums or YouTube more heavily. So you will see two completely different pictures depending on what your site's audience would be.
That is why the statistics can be misleading.
There's a long time during which Firefox was somewhat slow, and I remember the spidermokey team releasinf the famous Are we fast yet website that was saying no during this period.
I was so happy when Chrome came out. Over the years I've tried going back to Firefox and I've gone back to either Chrome, Ungoogled Chromium or Brave.
On Linux though, and blocking ads.
At the time 3.0 came out I was still in University and every other student I spoke to had the same experience. People also experienced in my first place of work in 2009.
This problem was ultimately fixed with 4.0 when they did something different with how multi-tab worked.
I was in perf engineering at the time. we would switch between a handful of string concatenation methods every browser release. it wasn't much about real performance, but just shifting trade-offs in the jit. but google PR team was very good at running in front of the changes and pointing their overly optimized way to magazines. so they would run an array concat test that was much faster while being much slower in plus sign concat, but they often left that out. anyway, everyone drank the coolaid. 100% of the v8 performance over spider monkey was not attaching debuggers and dev tools. and sadly, mozilla had to follow. nowadays we are mostly back to square one (still some niceties from dalvik missing).
true performance improvement came much later than that.
No I am not. I remember this clearly and all my friends were complaining about it before chrome was released. I just checked the dates. Firefox 3 was released a whole year before Chrome.
I really don't appreciate it when people tell me that I have been swayed by some big company, when my friends and I were complaining about it before we even knew that Google had a browser.
Firefox used to just completely lock up. Wouldn't load a tab. Chrome didn't with the same number of tabs. I am not talking about JS perf speed or anything like that, I am talking about the browser just not locking up when using more than few tabs.
When browsing Firefox 3.0 would randomly not load a site when say about 10 tabs were in the window and hang for about 30 seconds. It was annoying. There was an obvious issue with how it handled the tab isolation (it didn't really that was the issue). IE 8 and Opera didn't do this.
It has nothing to do with my impression of Google Chrome because Google Chrome didn't exist at the time.
Firefox briefly dominated the web in between.
Chrome dominance is a result of Google wanting to control the web and its dominance in the ads and search areas.
Chrome was simply better than Firefox, Internet Explorer, Legacy Edge. I am not a Windows Admin, but Chrome also offered an MSI package whereas Firefox didn't bother until years later. So it was easy for IT to roll out Chrome as part of the standard corpo install image and not an option with Firefox.
As for web development it offered better JS debugging and had a decent phone emulator built in and this was back in 2014. I am sure I could also debug phones as well with some reverse proxy shenanigans via fiddler and some open source tooling I forgot the name of now.
But the trend for IE started before this and continued after it.
The longer answer is yes, absolutely.
To parody the situation: a consortium of bridge engineers is discussing building standards, but somehow they’ve been lumped together with every girl named Bridget and every young boy making toy bridges with blocks, and they all have voting rights, and the girls are insisting that bridges must sparkle, and the boys think every bridge should be able to support helicopters and diggers.
Also, IE4 was such a magnificent leap forward in the web that effectively enabled support for modern apps, which bought IE a ton of goodwill from me that didn’t wear off for a decade or so.
* Kind of, I was born in the 2000s
Oh, what the heck, it’s been 20 years. Vertex Pharmaceuticals: shame on you. In the mid-2000s you had very poor taste in browsers ;-)
Man, literally every time the web platform had to choose between the IE way and the Netscape way they made the wrong choice huh.
For my own part, I made sure my employer had plans to remove IE6 from our support list the day Google officially did the same in March 2010. The very next day, I started adding code to our site that complied with official standards and worked perfectly on every other browser, and removing all the compatibility hacks we'd deployed to make that pig render a screen correctly. It was incredibly liberating.
FWIW that app is still running to this day: https://resultview.q2labsolutions.com/resultview/logon/logon...
Vanilla JavaScript just works. Marvel at the circa 2001 Login button!
The most complex part was a dynamic query builder where you could pick columns and various kinds of filters. We could have gone to the server each time the user changed the query, but I found it a lot snappier to do it all with document.write().
For a while, JavaScript was shunned by a lot of web shops. Applets and Flash were the future! Then Google Maps came out and showed what you could really do, and JS became cool again.
So many hacks... Not to mention the IE 5.0.0 select api bug, or the later uncatchable error in IE8's JSON parser... those were some rough years.
The base OS is far less relevant than it used to be, so it’s far less relevant that Chrome is merely an inner OS.
While Chrome is big, Safari still has significant market share today.
It was the worst two years of my life.
YouTube on TVs is actually a web app that loads into a stripped down, custom webview. The YouTube team doesn't have the resources to implement many web APIs, so they implemented just what they needed.
The problem is that they can't reliably update Cobalt versions on TVs, they can't ask users to update, and they can't just break older TVs in the wild. So the YouTube on TV frontend (not YouTube TV the service) has to only use APIs they shipped like 10 years ago.
And because it takes so long for an old Cobalt version to go out of support, they don't invest in implementing new features because they wouldn't be usable anytime soon. 10 years ago I was in a meeting with them where they said they couldn't implement something because they wouldn't be able to use it for 5 years... They still haven't implemented it.
[1]: https://developers.google.com/youtube/cobalt
But YouTube is also a very complex app. Yes it "just" exists to play videos, but the app is so much more than a video player. Browsing, searching, comments, chat, playlists, YT Live, subscriptions, profiles, ratings... there's a lot there.
Perhaps they could start with just cutting down their bloated 100x-duplicated 4MB CSS file?
However, it's clear that the devs are mainly composed of trendchasing sheeple who have drunk the Goog-Aid and are addicted to newness and reinventing wheels to make them square... because they have to justify their existence.
I call that trailer park logic:
They say: "Why go to college? That will take four years and I need a job now!"
Then four years later, while still in a dead end job: "Why go to college? That will take four years and I need a job now!"
Sometimes the best available move is physically escaping, just getting up and walking away and continuing until you find anything better than where you were at. For some weird reason, that move feels like giving up to people, until they actually do it and it works. This move is sometimes appropriate for jobs, relationships, addictions, violent circumstances, toxic social groups, politics, and so on.
If you've got next to nothing, then you have almost nothing to lose, and that can be a profound amount of freedom if it's seized. It's not always the right move, but sometimes the only winning move is not to play. Go find a better game.
It's also not unreasonable to believe that the two things are linked.
Why not call it woman logic? They are famous for strictly using long term planning and cold logic to plan their lives to the point they even joke about it.
You could've subbed in just about any nationality.
But you chose trailer park because the point was to pick a group of people that a bunch of other educated white collar people (I think the trailer park people would use the term "coastal elites" for this group, lol) like feeling better than, hence why the other groups wouldn't do.
If you wanted to make if harmless you could've chose any manner of public personality (politicians are gold mines for peddling short sighted stuff, plenty of examples to choose from) to name it after.
I started calling it that when I lived in the trailer park.
> If you wanted to make if harmless
It's already harmless. The people who live in trailer parks don't need your protection, and acknowledging that most of them don't want to be there isn't hurting anyone at all.
You're not acknowledging that they don't want to be there. You're assigning them a lack of long-term planning and ability to reason about how to achieve the best outcome for life. You're calling them "stupid" without using the word.
The comparison didn't include someone who chooses to take night classes at a junior college with hopes of transferring credits to another institution for an accredited degree, for example. No, they're too hopeless to find a third way. Having a background living in a trailer park in no way absolves you of dumping on people who still do in that comment.
I realize that I'm being hyperbolic to make a point and that life isn't simple when living in poverty. But I've been destitute as well, and I don't blame people who have trouble overcoming the same.
I told you what I meant. You're trying to tell me that it's not what I meant. As the the guy who said it i can tell you that you are wrong about my intent.
> You're calling them "stupid" without using the word.
I'm willing to own that it's Blunt language that might hurt feelings if you really want it to, but that wasn't the intent. I refuse to walk on eggshells because someone somewhere in the world might find some meaning i never intended if they really look for it.
As a human being sometimes I can't accurately communicate my full intent to everyone. However, as a normal human being you also have he obligation to try to see the best version of others peoples words before assuming the worst version.
Maybe we both failed a little here.
This is more like: "Why implement it? That won't be seen for five years and I need a promotion now!"
This way they could keep an old html/css/js implementation running alongside the upgraded one.
What new features?
The only "new features" Youtube implements is shoving shorts down your throat and taking five seconds to show video times on thumbnails despite the fact that the data is already there.
There's nothing Youtube requires from "new features" that can't be implemented in a browser tech from 15 years ago.
Also, Youtube the site doesn't have to deal with Cobalt-the-TV-app just like it doesn't have to deal with YouTube-the-mobile-app
Video encodings themselves are separate the client always selects the most favorable one from the available set (e. G., vp9 over av1 when hw decode for av1 is not present)
It's not proprietary
Annual revenue is a few dozen billions.
So what's the problem, here?
Minimizing developer pain is not a business objective.
It's a site that shows a bunch of text, a few images, and then loads and plays video. What features does it need on a TV that it's so hard to implement?
Edit: I know which ones, and they have very little to do with Cobalt, but with the fact that even high end TVs are often worse than a Raspberry Pi, and can stick around for a decade. But this is nothing ditching Cobalt would fix.
E.g. you can't run 4k video on some models that can technically show it because there's not enough CPU and RAM to run the browser, the video, the decoder, and the DRM at the same time, the video stutters
YouTube was perfectly usable 15 years ago, on the machines and software of the time.
I'll stop at the imminent conclusion that having Cobalt is a good thing for various reasons.
Looks like stable API like that is a perfect candidate for usage in an open source app that replaces the official one.
The significant shift IMO was when Windows 7 machines replaced the ageing XP machines. That is what I saw in the google analytics on the sites I was supporting at the time.
As an aside. IE7 was IMO worse in some ways the IE6. It had many of the same rendering bugs but was more subtle in how it failed.
And if you are actually concerned about the future of the web, instead of it's past, I would more concerned with Apple holding back development on Safari, to make people focus on writing native apps for mobile. There are so many apps that in the past would be websites, but end up being natively coded because that's the only way to get a good customer experience in mobile.
>I would more concerned with Apple holding back development on Safari
I think they are more concerned with the future of the web due to Google and so am I.
If Apple gets forced to allow other browser engines on iOS, it's game over. Google wins the web forever.
This has been annoying me personally because sometimes I'll "bookmark" something I want to come back to like a movie or a video game by Googling it and leaving the tab open on my phone's browser.
But now, seemingly when those pages are suspended, Javascript isn't allowed to run in them, so all my Google search tab thumbnails are just a static screen telling me that I need to enable Javascript.
Man, I love these tales of people doing the right thing cutting through the red tape.
A conspiracy to kill IE6 (2019) - https://news.ycombinator.com/item?id=39294406 - Feb 2024 (106 comments)
A Conspiracy to Kill IE6 - https://news.ycombinator.com/item?id=38210439 - Nov 2023 (1 comment)
A Conspiracy to Kill IE6 (2019) - https://news.ycombinator.com/item?id=28725293 - Oct 2021 (80 comments)
A Conspiracy to Kill IE6 at YouTube - https://news.ycombinator.com/item?id=28655890 - Sept 2021 (2 comments)
A Conspiracy to Kill IE6 - https://news.ycombinator.com/item?id=19798678 - May 2019 (363 comments)
The writing is on the wall.
Maybe we need to hire construction teams to break into peoples' houses and change them every 5 seconds.
Internet Explorer 6 was an incredible waste of resources. I developed primarily on a Mac OS system at the time, which was somewhat progressive in the industry, but in order to verify the functionality we had was working correctly on Internet Explorer 6 (which we still had observed was greater than 50% of the market share) I had to keep a PC on my desk just for IE6 testing.
There were a number of hacks that we could incorporate into additional override style sheets like conditional HTML comments that you could use to incorporate IE6 overrides or weird patterns that you could do by using asterisks that would allow you to target it specifically.
We didn't necessarily prioritize feature parity with IE6, but the site had to load and render correctly and support the cause of marketing the property that we were tasked to do. Once the adoption of it finally slowed, it was a great sigh of relief to the industry, and it made it feel like we could do anything we wanted to because we had been making concessions to it for so long.
It took a bit of diligence, but while I was there the release app payload never got over 400kb of initial JS, which for a modern React+MUI app is pretty good. Having to yank moment.js and a couple other libraries a couple times was not the most fun sset of conversations to have. Not to mention, replacing massive charting libraries with plain svg generation in React.
It bugs me to no end when developers don't seem to actually care about their craft at all.
But building the software wasn't enough, they used some scammy browser toolbar company (one of our competitors) to deploy this software silently and without any user intervention, all of a sudden millions of users overnight switched to chrome. It was deployed as a proxy botnet and Google knew full well what was happening. I sent a note to the humans at Firefox because we had a top 10 extension at the time and were in the midst of porting it to Chrome. They called their contacts and sure enough our suspicions were correct.
Google would later go on to buy that company because they were pushing so much traffic to Google's ad partners (Ad Meld being another acquisition).
We got screwed and were never able to recover from the run-around. I became friends with the folks on the Chromium team and we talked about how google used a botnet to launch Chrome over beers in a SF dive bar.
I still remember the time when people cherished the arrival of IE5.5 and IE6 later. They were once the best browsers.
Whoever this Croatian guy is, thankyou! True hero of the internet.
As soon as that banner popped up on Youtube we were able to tell our customers the same thing.
2 more comments available on Hacker News