Re: [Ohttp] Discovery

["Livingood, Jason" <[e-mail]>]

This is an interesting discussion over whether the WG charter should permit the WG to delve into operational implications in some manner.  In my mind these are as much operational implications as they are design choices or tradeoffs in the design - so putting aside the "operational implication" label I don't know why we'd not directly address what problems OHTTP is meant to solve and what the implications of the design choices may mean without making a judgement on whether the implications or tradeoffs are "good" or "bad".

In my reading of the discussion & charter, it seems like if the design is intended to prevent revealing client source IP addresses to destination HTTP servers then that might mean some of the existing mechanisms for content localization (e.g. based on geolocation & network) would not work or would be less effective - and this may be totally fine for some use cases & threat models. And perhaps new mechanisms will arise to solve this, that provides some sense of geographic and network location but not at too specific a level. But in using OHTTP, ISTM the user or user-agent is making a tradeoff between the pros & cons of sharing that source information (e.g. CDN performance) vs. the privacy pros & cons of that source information being available to a web app platform. So why not explain that?

As well, is it worth including somewhere an explanation that within HTTP content exchange & user agents that cookies or authentication can also be used for user tracking and that OHTTP would not protect against tracking via those mechanisms? Those mechanisms may even be more central to how user tracking occurs.

Jason