3 ms·
I think the bad state to be in is: * There is a number of implementations of the same platform. * Those implementations are competing and the market encourage
by munificent 8y ago
I think the bad state to be in is:
* There is a number of implementations of the same platform.
* Those implementations are competing and the market encourages them to be in-compatible with each other.
You can avoid that bad state by having an explicit standard and applying pressure on all implementations to meet that standard. It usually comes at the expense of slower innovation.
But another viable way to avoid that bad state is to simply not have competing implementations. A single canonical implementation also solves the problem of ensuring all users get a compatible experience. It can come at the expense of evolving in a way that doesn't meet user needs because there isn't a competitive incentive to win users.
I don't think there are perfect solutions, but I also don't think it's the end of the world if this becomes a monoculture. There are tons of "platforms" that are effectively mono-cultures and seem to be OK. Every rechargeable tool company has its own battery pack form factor. Up until recently, each laptop company had a different port for the AC adapter.
What I think this is really showing is that the market size of the web is shrinking relative to the size of the browser standards. Mobile apps have eaten up so much user share and HTML+CSS+JS+etc. has gotten so big and complex that the desktop web market can't effectively support multiple independent browser implementations any more. It's just too much work for too little return.
There's maybe an interesting lesson here in not letting your platform get too complex. New features are always nice, but they have a cost. If you pile on too many of them, you may undermine your platform's ability to support multiple independent implementations.
- pcwalton 8y agoSo whether the Web can run on Rust-based parallel engine components should be beholden to the reporting structure at Google? Let's be clear: that is the future you are advocating.
- Yoric 8y ago...or whether the web can/should support any technology that could possibly threaten Google's dominance.
- Yoric 8y agoAlso, congrats for Rust 2018 :)
- tree_of_item 8y agoIs Microsoft not capable of adding whatever they like to Chromium? I don't understand.
- Yoric 8y agoWhy would they be? The owner of the repo is Google. Google gets last word on anything Microsoft wants to include in Chromium/Edge.
- dragonwriter 8y ago> The owner of the repo is Google That has no impact on what MS can do downstream as long as it's an open license. If Microsoft, or Opera, or anyone else downstream wants to pull out parts of Chromium and replace them with Rust components (including a parallel renderer), they are free to do so. It increases their maintenance cost to maintain the unique components, but so would maintaining a whole separate from-scratch project instead of being downstream.
- camgunz 8y agoYour comment isn't too far away from accusing GP of shilling. I think probably it would be more productive to discuss any of the points he made about platform complexity or technology homo/heteroculture without going full Mozilla v. Google here.
- munificent 8y agoWow, easy there. I'm guessing this is a stressful day around the Mozilla watercooler. This is bad news for Mozilla. Web authors will target the behavior supported by a majority of user browsers. With many independent browsers, there is no implementation majority, just a plurality. Majority behavior only comes from a standard and all implementers are incentivized to work with that standard. When there's only a few, it's possible for a single one to become the de facto "standard". The web is effectively moving to a first-past-the-post election. Because Microsoft is adopting Chromium, Mozilla's engine and Safari may very quickly become minority ones. This might suck for users, but that's not entirely clear. Obviously, in a perfect world, there would be infinite engineers building infinite implementations of every platform. In reality, every engineer-hour spent working on, say, a new implementation of COBOL is an hour not spent on software that might impact more users' lives in more important ways. Maybe a couple of commodity web browsers and engineers working on other more important stuff is better for society? How many implementations of CSS does the world really need? <shrug> Either way, I'm not advocating anything. I'm more interested in understanding what are the causes and effects — both bad and good — of a platform going from three implementations to two. I'm approaching this as sociology, not as someone who has skin in the game. I do work for Google and did work as part of the Chrome org, but I don't have enough expertise to make any claims about whether this is an overall good or bad thing for the world. I'm just interested in all of the consequences.
- pcwalton 8y ago> Maybe a couple of commodity web browsers and engineers working on other more important stuff is better for society? How many implementations of CSS does the world really need? Shouldn't users have the best possible CSS engine? If there's only one, and it's beholden to the reporting structure at Google, then disruptive innovations are less likely to happen. People who don't work at Google should be allowed to develop the Web platform, without asking Google for permission.
- munificent 8y ago> Shouldn't users have the best possible CSS engine? Shouldn't they have the best possible COBOL compiler? The best possible VRML renderer? The best possible ICQ client? Obviously CSS is way less dead than those, but there is always an opportunity cost. It's not a valid engineering argument to say "we should spend resources on X" without considering what else those resources could be spent on. My initial comment is really just observing that maybe Microsoft's move actually does imply that they believe, no, the world doesn't need another CSS engine. I don't know if that's true or not, but I think it's interesting to ask the question. We sort of a tacitly assume that the web will grow and grow forever and ever. But maybe that's just because we've only seen the first half of its life cycle. Maybe it is reaching a plateau and becoming a commodity. There are good questions about whether that's true and, if so, whether that's a good thing. But I don't think it's illuminating to just assume the best way to improve the world is to have as many browser implementations as possible. Like an entire Dyson sphere populated solely by engineers each writing their own HTML parser. > then disruptive innovations are less likely to happen. Maybe the disruptive innovation has happened and the disruption was to move off the web. The Internet and HTTP is doing great. What mobile app isn't on the Internet? Maybe HTML+CSS+JS is no longer the optimal user interface language for it. > People who don't work at Google should be allowed to develop the Web platform, without asking Google for permission. I guess I don't see what you're getting at. Browsers don't write themselves, so if you don't want to ask one of a couple of giant rich corporations to do something, that means you better have deep pockets yourself. Even if Microsoft kept their web engine, how does that help? Now instead of asking Google, Mozilla, and WebKit for permission, you have to ask Microsoft, Google, Mozilla, and WebKit for permission and then get them to all agree on it.