What if Chrome broke features of the web and Google forgot to tell anyone? Oh wait, that's exactly what happened

Browser monoculture bad


Feature "Browser monoculture" is often bemoaned as a threat to the web. According to Statscounter, which tracks browser use, over 70 per cent of the market is made up of people using Google Chrome or another browser based on the underlying Chromium project.

What web advocates worry about when they say this is bad is that Google can effectively determine the future of the web by determining which features to support and which not to. That's a lot of power for a single company that also has an effective monopoly on search and advertising.

What would happen if Chrome decided to break fundamental features of the web and didn't even feel the need to tell the web's very developers?

Well, we can answer that question because that's what Chrome did.

Earlier this year, it emerged Chrome engineers had decided the browser should no longer support JavaScript dialogs and alert windows when they're called by third-party iframes.

That means that if something is embedded from another website, let's say a YouTube video, Chrome wants to stop allowing that embedded content to call the JavaScript alert function, which opens a small alert window. Eventually Chrome aims to get rid of alert windows altogether.

The wheels were set in motion for this development in 2020, starting with an entry in a bug tracker that Chromium and other browser engineers read. The proposal was shared with the Web Hypertext Application Technology Working Group (WHATWG), which sets today's HTML standards. No one else spots it.

Then a Chrome developer happens to mention it in passing on Twitter in August this year. That raises a riot of angry web developers, in the face of which Chrome postpones the move until January 2022, when it will again try to remove the features and hope that this time no one notices.

You know what isn't happening here? No substantial public discussion happens, certainly not with builders of websites. No announcements.

Google puts its idea forward on its own mailing list, programmers at Apple working on Safari's WebKit and at Mozilla working on Firefox are invited to agree with the change on WHATWG's GitHub page, and they do so with plans to implement the functionality. The HTML standards are updated. Google gets what it wants and the web breaks.

"Big company bad" is hardly news at this point, especially if the big company is Google with its browser market dominance, but there's more going on here than that and it's worth picking apart a little.

Discontinuing major features is rare. Part of what's amazing about the web is that you can still go to the very first web page and view it in any browser. The web is the web in large part because of this high level of backwards compatibility. To their credit, browser makers have generally been very good about making sure changes don't break the web.

That said, change happens. Most browsers don't support the blink tag anymore. Try using applet or AppCache – both are gone. That is, they're gone from the official web standard. Individual browsers may still support them, but they are no longer valid HTML.

It is crucial that decisions about what should and should not be part of HTML happen are not only taken by standards bodies, previously the W3C and now WHATWG, but that those decisions should come after lengthy public discussion and testing.

The other key problem here is the lack of communication beyond the close-knit community of browser engineers. When the web developer community finds out Google is going to break a ton of websites through a tweet, you know communication has failed. Clearly no wider consultation took place when it ought to have done for a change like this.

But there was a follow-up tweet that's actually far more disturbing than the news of alert() disappearing.

The tweet comes from Chrome software engineer and manager Emily Stark, who is of course speaking for herself, not Chrome, but it seems safe to assume that this thinking is prevalent at Google. She writes: "Breaking changes happen often on the web, and as a developer it's good practice to test against early release channels of major browsers to learn about any compatibility issues upfront."

First, she is flat out wrong – breaking changes happen very rarely on the web and, as noted, there should be a process for making sure they go smoothly and are worth the "cost" of breaking things. But second, and far more disturbing, is the notion that web developers should be continually testing their websites against early releases of major browsers.

That's actually why there needs to be stable, well-communicated web standards – so developers don't have to do ridiculous things like continually test their websites to make sure they're still working. You build the site using the agreed-upon standard and it works as long as the web does. Full stop. That is the point of standards. If the standard has to change, tell people about it. Don't wait for them to notice it in a development build. That someone of considerable stature in the Chrome project would think otherwise should be a red flag.

Web developer and advocate Jeremy Keith points out something else that's wrong with this idea. "There was an unspoken assumption that the web is built by professional web developers," he writes. "That gave me a cold chill."

What's chilling about the assumption is just that, it's assumed. The idea that there might be someone sitting right now writing their first tentative lines of HTML so that they can launch a webpage dedicated to ostriches is not even considered.

What we are forced to assume in turn is that Chrome is built by the professional developers working for an ad agency with the primary goal of building a web browser that serves the needs of other professional developers working for the ad agency's prospective clients.

As Keith points out, this assumption that everyone is a professional fits the currently popular narrative of web development, which is that "web development has become more complex; so complex, in fact, that only an elite priesthood are capable of making websites today."

That is, as Keith puts it, "absolute bollocks."

I've been teaching people to build things on the web (in one form or another) for almost 20 years now, and you know what? It's no harder to write HTML now than it was 20 years ago. There's no more need for the supposed complexity of the modern web than there ever was. In fact, I think it's actually the opposite.

I find myself increasingly turned off by sites that are so obviously overengineered. I've started to notice the beautiful simplicity of an HTML page. Just the simple fact that it loads without a spinning circle makes it stand out on the web today.

You'd be forgiven for thinking that the most common content on the web these days is that little spinning circle you see while you wait for simple text content to pass through several layers of unnecessary complexity before being seen.

The complexity of the modern web seems like the law of diminishing returns in action. Developers keep pouring on the JavaScript and we keeping getting... less of what we actually want.

That's not to say there isn't a time and place for complexity. Building turn-by-turn navigation with real-time map updates calls for some complex JavaScript and it's great that the modern web has the standards to make that possible. But not every webpage needs to be that. The web is not a place just for professional developers, it's a place anyone can build pretty darn near anything, and it certainly isn't a place where Chrome gets to dictate the tools we use or who can participate. The web is for everyone, not just developers.

Just a friendly reminder, Firefox is an excellent web browser. ®

Editor's note: This article was revised after publication to acknowledge Google's dialog box proposal was discussed on GitHub with WHATWG as per procedure, and that WebKit and Firefox developers supported the change and planned to implement it. Also, we are happy to clarify that Google did assess the estimated impact of its proposal: it claimed 0.00906 per cent of page loads would be affected by the change. Our central point that this process and outcome was poorly communicated to the wider web developer community still stands.

Similar topics


Other stories you might like

Biting the hand that feeds IT © 1998–2021