XDEFiANCE'e Quality Internet Shop

The bittersweet consequence of YouTube’s incredible growth is that so many stories will be lost underneath all of the layers of new paint. This is why I wanted to tell the story of how, ten years ago, a small team of web developers conspired to kill IE6 from inside YouTube and got away with it.

We somehow got away with our plan to kill IE6 without facing any meaningful corrective action. Few people even knew we were involved at all and those that did, did not want to bring attention to it or risk encouraging similar behavior. At a beer garden in San Francisco, our boss, winking his hardest, made us swear to never do anything like this again. We agreed, toasted IE6 falling into single digit percentages, and never snuck anything into production again.

The plan was very simple. We would put a small banner above the video player that would only show up for IE6 users. It would read “We will be phasing out support for your browser soon. Please upgrade to one of these more modern browsers.” Next to the text would be links to the current versions of the major browsers, including Chrome, Firefox, IE8 and eventually, Opera. The text was intentionally vague and the timeline left completely undefined. We hoped that it was threatening enough to motivate end users to upgrade without forcing us to commit to any actual deprecation plan. Users would have the ability to close out this warning if they wanted to ignore it or deal with it later. The code was designed to be as subtle as possible so that it would not catch the attention of anyone monitoring our checkins. Nobody except the web development team used IE6 with any real regularity, so we knew it was unlikely anyone would notice our banner appear in the staging environment. We even delayed having the text translated for international users so that a translator asking for additional context could not inadvertently surface what we were doing. Next, we just needed a way to slip the code into production without anyone catching on.

To cement their authority over the YouTube codebase during the integration into Google, the early engineers created a specialized permission set called “OldTuber”. OldTuber granted you the ability to completely bypass the new Google-oriented code enforcement policies, enabling anyone holding it to commit code directly to the YouTube codebase, with only the most glancing of code reviews from anyone. No need for code readability. No need for exhaustive tests. No need for maintaining code coverage. If you broke the site by improperly wielding OldTuber status, it was on your head and you would lose the privilege immediately, if not your job. So you just had to be a good citizen and never break the site. Our boss, an early YouTube engineer himself, had gone out of his way to ingratiate the web development team with the rest of the early YouTube engineers. Through his efforts, a couple of us eventually found ourselves in possession of OldTuber status, despite never having been a part of the original team. It was like we were just walking down the street when someone mistook us for valets and handed us the keys to their Ferrari. For better or for worse, we were not exactly the types to just hand the keys back and walk away. We saw an opportunity in front of us to permanently cripple IE6 that we might never get again. If this went at all wrong, a number of us would surely be fired. Our most renegade web developer, an otherwise soft-spoken Croatian guy, insisted on checking in the code under his name, as a badge of personal honor, and the rest of us leveraged our OldTuber status to approve the code review. The code merged into production and our banner went live a few days later.

Eventually the YouTube engineering management did ask themselves how the decision to deprecate IE6 was ultimately made, given it happened so quickly and seemed conspicuously premature for a media site of our scale and with such a wide user base. Once they realized what had happened, they cornered our boss for details, grappled with the consequences of our actions and begrudgingly arrived at the conclusion that the ends had justified the means. Between YouTube, Google Docs, and several other Google properties posting IE6 banners, Google had given permission to every other site on the web to add their own. IE6 banners suddenly started appearing everywhere. Within one month, our YouTube IE6 user base was cut in half and over 10% of global IE6 traffic had dropped off while all other browsers increased in corresponding amounts. The results were better than our web development team had ever intended.

The Historical Share of Different IE Browser Versions (https://www.w3counter.com/trends)

It turned out that a handful of us had entered YouTube at an interesting time… several months after YouTube had been acquired by Google but before Google had begun deeply integrating YouTube into their larger organization. The early YouTube engineers were rightfully territorial and initially hesitant to adapt to Google’s infrastructure and norms. With their penchants for gray-hat hacking, fast cars, and hard whiskey and an uncommon number of piercings, tattoos, and minor arrest records, many had been rejected during previous Google interviews. Ending up at YouTube instead, they found themselves breaking their backs to stay ahead of exponentially growing traffic while having to constantly defend against critics explaining how Google Video would imminently kill them. By the time they were acquired into Google, many of these engineers had come to view their outcast identity as a critical component of their eventual success.

Now read this Page Weight Matters Three years ago, while I was a web developer at YouTube, one of the senior engineers began a rant about the page weight of the video watch page being far too large. The page had ballooned to as high as 1.2MB and dozens of requests. This… Continue →

Three years ago, while I was a web developer at YouTube, one of the senior engineers began a rant about the page weight of the video watch page being far too large. The page had ballooned to as high as 1.2MB and dozens of requests. This… Continue →