Pai makes a lot of questionable arguments:

The FCC’s regulations weren’t about protecting consumers’ privacy. They were about government picking winners and losers in the marketplace. If two online companies have access to the same data about your Internet usage, why should the federal government give one company greater leeway to use it than the other?

1. Is he implying that the previous FCC was intentionally boosting content providers over ISPs? What motivation would they even have for doing this?

2. Content providers and ISPs do not directly compete with each other in the course of their primary business, so in what respect does the rule create winners and losers in the marketplace?

I suppose he could mean in the advertising marketplace. I guess I would prefer that both Facebook and Comcast be losers in that marketplace though.

3. Two online companies do not have access to the same data. Facebook has data + metadata for any interaction you have with their platform. Your ISP has metadata for interactions you have with any platform, along with data for any plaintext interactions. This is a massive difference in scale.

Users have the choice not to use Facebook, and thus not provide them with data. Users do not have the choice to access the internet without making use of a public utility (i.e. their ISP).

He cites an expert saying:

Rather, the most commercially valuable information about online users . . . is coming from other contexts,” such as social-media interactions and search terms.

I can see how user-shared data might be considered more commercially valuable, but this misses the point about the differences in the kinds of data we are talking about. When someone publishes on social media, they're making an explicit choice to share information with the public.

When that same person visits a website, it is not commonly understood that this act might reveal personal information (though end-users ought to be more aware of this). Consider a teenager doing research on pregnancy tests -- they do not intentionally publish information, but the metadata (webpages visited, time of visits, etc) is potentially revealing.