It is currently widely reported that the US House of representatives has collected on the order of 1.3 million documents as part of its current antitrust investigation of large tech companies.

Is there a way to understand how congress actually processes and extracts information from one million documents? Of course it's not possible for a committee to read all of them, but do they use something like antitrust detection software to cross-correlate them all and look for patterns, or pay a lot of money to a consultancy to process it in a proprietary way, or something else?

1 Answer 1


Realistically, the same way that any law firm deals with large data dumps as part of the discovery process. There are a variety of software products that allow you to load and index a bunch of documents so that you can easily search them. Someone defines some key search terms (i.e. "App Store + Prime Video" if you're looking for anything relating to the deal Apple and Amazon negotiated on Prime Video) and then a small army of staffers/ lawyers does the tedious work of reading the documents that come up to see if they contain anything interesting.

Depending on who is doing the counting, 1.3 million documents may be a rather large over-estimation. Each email is technically a different document. So if you have an email thread that goes back and forth between Executive A and Executive B a total of 10 times, you're going to have 10 Executive A documents and 10 Executive B documents for a total of 20 documents but all the information would be in the last email (assuming that everyone just replied each time rather than editing the history). When you get strategy documents that are copied to dozens of people, that can pretty quickly generate dozens or hundreds of duplicate documents for many documents in the data dump.

  • I see, Thanks! After reading this, here's what I'm wondering (I don't know if it's too late to add to the question); a search engine and a competent operator could find things like "Shall we do some colluding next week?" quickly, but might struggle with finding "I would like you to do us a favor" if author were smart and left out the "though". Do these discovery aids include any next level or AI features that are able to flag potentially important exchanges or statements that a simple search engine would miss? Then again, is that better asked separately in for example Law SE, or AI SE?
    – uhoh
    Commented Aug 1, 2020 at 8:15
  • I'm just thinking that these days tech people know how email works and as it was put so well, "the internet is written in ink" and so may simply avoid easily searched terms.
    – uhoh
    Commented Aug 1, 2020 at 8:18
  • 5
    @uhoh - Last time I had any peripheral interaction with them, no. The AI was a bunch of underpaid lawyers reading through anything that was flagged as moderately interesting (i.e. anything from an exec of company A to an exec of company B). But times may have changed. On the other hand, never underestimate the stupidity of executives sending email. Even though on one level everyone understands that corporate emails are hardly private, people get so accustomed to email feeling private that they say stupid stuff all the time. Much to the chagrin of the legal department. Commented Aug 1, 2020 at 11:29
  • Re executives writing stupid things in emails: See for example Barclays and their "craptacular loans."
    – Kevin
    Commented Aug 3, 2020 at 21:13

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .