Document should discourage features that'd lead to a more centralized web

[pes10k]

This came up during discussion on the Privacy Principals document, w/r/t how some privacy-improving proposals might in practice end up giving further advantage to large companies and orgs. That group agreed that concern would be a better fit here.

https://github.com/w3ctag/privacy-principles/pull/293

Note I've also briefly discussed adding this concern to the societal-impact-questionnaire with @hober , in addition to (or instead of) adding it here

[hadleybeeman]

Hi all. We thought carefully about this topic when we wrote the first draft of the EWP, and — as you can imagine — it has come up a lot in the ensuing revisions.

We have written each principle to be about the ethics, with the technologies in the explaining text. The ethical issue that underpins centralisation is the power imbalance: when pieces of the web are centralised, it means that individuals or small parties lose power in comparison with the centralising power.

We wrote that up in 2.8 The web must enhance individuals' control and power:

We recognize that web technologies can be used to manipulate and deceive people, complicate isolation, and encourage addictive behaviors. We seek to mitigate against these potential abuses and patterns when creating new technologies and platforms, and avoid introducing technologies that increase the chance of people being harmed in this way. As part of this, we favor a decentralized web architecture that minimizes single points of failure and single points of control. We will also build web technologies for individual developers as well for developers at large companies and organizations. The web should enable do-it-yourself developers.

We also wrote that we favor a decentralised web architecture, instead of saying that we will only accept one, because — as @mnot points out on the PR — DNS is relatively centralized and yet is a pretty functional part of how things work right now. We aren't intending to condemn it.

@pes10k, with all that in mind, if you would like to try to strengthen the text in principle 2.8, we would appreciate any suggestions.

In the meantime, we are closing PR 95.

[pes10k]

hi @hadleybeeman ,

I appreciate that there is some overlap here, but before i suggest new text I want to better understand the TAGs thinking with the existing 2.8.

My concern (coming from a relatively small browser vendor) is that the Web platform should avoid features that, by design, work more poorly in browsers with smaller number of users than in browsers with larger number of users. As one example, the platform should (to the degree possible) not include features that create privacy protections in large scale data collection through noise injection. I worry that features with this structure would create (even more) incentives for sites to encourage visitors to use dominant browsers over small browsers (among other bad things).

I appreciate that my clumsy wording might have done a bad job distinguishing this kind of feature from DNS, but they seem meaningfully and significantly distinct to me.

Im trying to get a sense of whether the TAG thinks:

1. the above concern is already addressed by 2.8
2. the above concern should be incorporated into 2.8, or
3. the above concern doesn't need to be addressed in the Ethical Web Principals document?

Just trying to make sure I understand before i start bothering ya'll with more text to review

[mnot]

@pes10k your statement:

the Web platform should avoid features that, by design, work more poorly in browsers with smaller number of users than in browsers with larger number of users

concerns me. Yes, how these functions should be a consideration, but a blanket 'should avoid' is too strong -- these issues are very nuanced and need to be evaluated in context.

[pes10k]

@mnot grateful for help word smithing, and I understand your concern.

Im just trying to get a broad sense of whether folks share the concerns and (if so) where/how they should be brought into the document. Then very happy to start improving wording and crafting text

I understand you to be saying "some form of this concern would be good to incorporate, though not as currently stated". Is that accurate?

[mnot]

I'm a bit more skeptical than that. In some cases the privacy gains may be worth the centralization risk, so I'm leery of establishing any principle (which this document is about) here. It's more of a factor to consider.

[pes10k]

I see. I'm sure that there is some scenarios where privacy gains are worth centralization risk, but I bet we disagree on whether such situations are so common that they'd nullify the general principal.

Principals dont exist in a vacuum, they're often in tension with each other, are about broad goals and don't (necessarily, in and of themselves) imply specific decisions or how to resolve trade offs, etc etc etc. This is true for many of the existing Ethical Web Principles as well.

But "the web should not adopt features that encourage centralization" seems like an important principal for an open system, even if that principal is sometimes in tension with other important principals/goals.

Grateful for others' thoughts too

[pes10k]

Sorry, one last though before i pause for a bit, and maybe this will help close the gap a bit between us.

But a further reason i think a principal is important here is to not say "when you have to trade less-centralization against more-privacy, you should prefer less-centralization".

My goal instead is to say "if a feature can only be implemented in a privacy-respecting way by centralizing the Web platform, that is a reason, on balance, to be skeptical that the feature is good for the Web platform"

[mnot]

That's more reasonable, yes.

[pes10k]

Hi TAG members, I wanted to ping you all on this again.

I appreciate the decision to not block the Ethical Web Principals document on this issue, but I think its a topic thats urgent to have the TAG's position and guidance on in some document, given the speed and expected impact on a number of proposals being worked on that could have significant centralization impact (mostly thinking of recent private advertising work, but other areas too.)

So if not the Ethical Web Principals document, would the TAG be willing to include guidance in the Web Design Principals document?