Somehow I don't think they would do this without their architecture being based on a pre-existing lower-resolution backup layer to maintain information and support. It's not like the model needs every piece of information coming out of Facebook users - they only need the important stuff and could conceivably delete unnecessary information and not lose anything. Given the massive redundancy coming out of thousands or millions of similar people, the global patterns won't change when some subset of users deletes their information. Even local patterns should be filled in by a fresh influx of data fairly quickly.
The unique views problem is interesting from a marketing perspective though. I doubt it will change much but for larger companies it might require adjustment.
Also interesting that they are allowing a theoretical cap on the information they collect. I imagine they'll be moving more into services to avoid future data privacy issues. That means B2B too, FYI...
The unique views problem is interesting from a marketing perspective though. I doubt it will change much but for larger companies it might require adjustment.
Also interesting that they are allowing a theoretical cap on the information they collect. I imagine they'll be moving more into services to avoid future data privacy issues. That means B2B too, FYI...