On PPK’s moratorium on new browser features

Famed developer and commentator Peter Paul Koch (PPK) recently called for “a moratorium on new browser features for about a year or so”. If you haven’t read his article Stop pushing the web forward, give it a look; he raises interesting points, as he always does.

(Let us say now: we’re all big fans of PPK; he’s undergone personal attacks for his article — we’re going to disagree with his central thesis, while continuing to love him deeply, and thanking him for starting this discussion.)

In many ways, we Opera devrel folks feel his pain. Each of us is familiar with the feeling of coming back from a vacation and not understanding the Twitter conversations about specs that were launched in the fortnight we were lounging next to a pool / winning hearts with our bachata / touring ancient ruins / having it large in Magaluf / doing emoji poos and wishing we hadn’t eaten so many U+1F364s last night. (Delete as appropriate depending if you’re Bruce, Shwetank, Mathias, Vadim, or Andreas.)

There’s a lot to learn, and the web platform is becoming more complex. Even Ian Hickson, the editor of HTML5, said:

The platform has been too complex for any one person to fully understand for a long time already. Heck, there are parts of the Web platform that I haven’t even tried to grok — for example, WebGL or IndexDB — and parts that I continually find to be incredibly complicated despite my efforts at understanding them

…and that was two and a half years ago!

But it’s not necessary to remember the minutiae of every specification. It’s necessary to know what’s possible, and have access to a search engine to find the spec or tutorials to find details. How many of us memorise the syntax for CSS gradients, or remember every piece of the Web Audio API syntax? But that doesn’t stop us using them when we need to.

But PPK’s complaint isn’t primarily about complexity:

We should focus on the web’s strengths: simplicity, URLs and reach. The innovation machine is running at full speed in the wrong direction.

PPK cites an example:

To me, Navigation Transitions exemplifies what’s wrong with new browser features today. Its purpose is to allow for a smooth transition from one web page to another, to the point of synchronising the animations on the source and destination pages […] We’ve done without for years. More importantly, end users have done without for years, and are quite used to a slight delay when they load another page […] But why do web developers want navigation transitions? In order to emulate native apps, of course. To me, that’s not good enough.

But the point here is that users do want such things, because they’ve now become used to experiences available in native apps. And we know that consumers love the app experience; in April 2014, the mobile analytics firm Flurry reported

Apps continued to cement their lead, and commanded 86% of the average US mobile consumer’s time, or 2 hrs and 19 minutes per day. Time spent on the mobile web continued to decline and averaged just 14% of the US mobile consumer’s time, or 22 minutes per day.

Many of the new “features” coming to the web, like Service Worker or Installable Web Apps, are designed to enhance the web experience for end users — experiences they’ve become accustomed to from native apps but weren’t achievable previously on the web. That’s a win.

We also respectfully disagree with PPK that there’s a dichotomy between adding native-like user experiences and protecting the web’s core strengths of simplicity, URLs and reach.

Let’s address those core strengths in reverse order:

Reach

We’re reaching the point when some organisations are willing to forfeit the reach of the web because they want the features of native. For instance, Myntra does this — and its parent company Flipkart is planning to do so soon as well; they have disabled their site on mobile and urge people to use the app (although you can still use the site on desktop).

Uber’s service is only available through the app. Arguably, that’s understandable because sometimes the Geolocation API is not very accurate on web, and really accurate location info is critical to a taxi hailing service. But that’s an argument for making the Geolocation API better, rather than stopping development.

If we slow development of the web, we risk losing more services to native, thereby diminishing the web’s reach.

URLs

If you get a great big saucepan, and boil the web in it all weekend, periodically skimming off the scum of YouTube comments, porn and photographs of horrible kittens (tautology?), when you look under the lid on Sunday night you’ll find you’re left with URLs. (Or URIs. Or URNs. Who cares what the difference is?)

It’s called “the web” because it’s a network that joins resources together, and those resources are individually addressable.

Modern standards are designed to preserve URLs. Take Service Worker, for example; the explainer document for Navigation Controller (the previous name for what’s become Service Worker) says

It forces you to have URLs! Some modern apps platforms have forsaken this core principle of the web and suffer for it. The web should never make the same mistake.

Similarly, the Web Manifest spec defines a web app’s start and scope in terms of good old-fashioned vitally-important URLs. The proposed Upgrade Insecure Requests spec tries to ensure that no links break if a developer upgrades their server to HTTPS in order to provide a better (more secure) user experience.

There’s a lot the web can learn from native (without slavishly emulating it), but linkability is something that native needs to emulate from the web; see the Rube Goldberg machine-like App Links proposal to see how Facebook is trying to bring “deep linking to content in your mobile app”. We have URLs; PPK is right that we need to jealously preserve them, so modern standards attempt to do just that.

Simplicity

There’s a deeper complexity to the modern web platform than the sheer volume of features. It’s to do with the way the specs were written, the timescale over which they’ve been written and the fact that some features that we rely on have never been specified at all.

One example is HTML5 Parsing. For years, developers had to deal with the different DOMs that browsers constructed from invalid markup (which, as we know, is the vast majority of the web). This was allowed because HTML 4 never specified what to do with bad markup, so browsers were free to do as they saw fit.

HTML5 changed that, and now all browsers worth shaking an angle bracket at produce the same DOM regardless of the validity of the markup. This has produced a huge boost in interoperability, benefitting consumers and saving developers megatons of heartache.

A more current example is XMLHttpRequest which was never formalised and standardised until years after Microsoft implemented it and everyone else reverse-engineered it and copied it.

XHR is hardly a beautiful API, and will be replaced by the Fetch Standard which aims to simplify and unify network requests. Its preface says

At a high level, fetching a resource is a fairly simple operation. A request goes in, a response comes out. The details of that operation are however quite involved and used to not be written down carefully and differ from one API to the next.

Numerous APIs provide the ability to fetch a resource, e.g. HTML’s img and script element, CSS’ cursor and list-style-image, the navigator.sendBeacon() and self.importScripts() JavaScript APIs. The Fetch Standard provides a unified architecture for these features so they are all consistent when it comes to various aspects of fetching, such as redirects and the CORS protocol.

Modern standards are all about explaining the platform to simplify development, and ensuring a solid, understandable foundation upon which to build.

This is built on a design philosophy called the Extensible Web Manifesto. It’s too much to explore here, but The Chair of the W3C Extensible Web Community Group, Brian Kardell, wrote us an article about it called Sex, Houdini and the Extensible Web.

Immediacy

A central pillar of the web that PPK doesn’t mention is what I call “immediacy”. When you make a change to a web site, the next visitor gets the updated version immediately. With native apps you have to publish to an App Store, your user is alerted that there’s an updated version and, when they have Wi-Fi, they’ll update it. Maybe.

Installable Web Apps give us an app-like experience — an icon on the homescreen, potentially working even while offline — but retain the immediacy of the web because the app is hosted on a server. In fact, the app is actually — wait for it — a web site with a URL pointed to by the homescreen icon. We combine the strengths of the web with the user experience of native.

Conclusion

At Opera, a lot our developers work on bringing the web to people who otherwise wouldn’t get it, either with Opera Mini, or by reducing Chromium’s memory consumption so that it works on the lower-specification devices that most of the world uses.

We know that the fastest growing mobile phone markets don’t use apps, so by artificially slowing the pace of evolution on the web, we’re deciding that these people should get a second-class online experience.

It’s imperative, we believe, for the web to continue to add new features — like Service Workers and Installable Web Apps, just as we added native video, the Audio API, the <picture> element, Storage APIs — that extend what the web can do so that it continues to grow and provide the reach that PPK wants, and that we want.

Opera welcomes new developments that make the web better for users. We’re the only browser manufacturer that isn’t also trying to sell an Operating System or locked-down device. So for us, it’s vital that the web continues to thrive — and we believe it’s vital for everybody.

This was written by Bruce Lawson, with input from the rest of Opera’s Developer Relations team. Disagree? Please, write a commentary post and tweet us the link!

Other people have been commenting, too:

18 August: PPK’s response Stop pushing redux was published.