The Meta antitrust trial entered Week 5 with Meta marketing head Alex Schultz on the stand again, before the FTC finally called its economist. Sparks flew.
Interesting approach from Hansen in his impeachment of Hemphill. It seems based on how you phrased it, that it was effective in undermining his credibility.
However, it does beg the question, if in fact Hemphill HAD prejudged the outcome of the case, does that in any way reduce the substantive merit of his testimony based on evidence other than his own opinion? Of course it does call into question the validity of Hemphill’s opinion. But if Hemphill’s testimony is primarily his interpretation of factual evidence, isn’t it just as likely that his interpretation is correct or not correct, independent of his views on the case as a whole?
I’ll be interested to see if the judge’s opinion reflects this cross.
I think the issue as Hansen framed it was that if Hemphill had an opinion before the case, he's not doing a neutral evaluation of the evidence and is disregarding evidence that doesn't fit with his preconceived notions. And if all his testimony boils down to is his own interpretation of fact evidence, that gets into the territory of testimony that's not helpful to the court, as experts can't just regurgitate what fact witnesses say or tell the court how it should weigh evidence. Hansen tried to get some answers in this direction, asking Hemphill if he knows better than Meta executives on these topics, is he calling Meta execs liars, etc.
For me, it was effective to see -- even more so in the stuff I found from 2020, rather than the 2019 "roadshow" deck -- that Hemphill was making sweeping conclusions about Facebook with very little to back it up at the time. That's not to say he can't back it up now. But it is telling that despite all that, he doesn't have an opinion on whether Instagram would have done better or worse without being acquired by Facebook. That's kind of the whole point. So if someone who is purportedly biased can't even agree with the FTC's theory on that score, how effective is the testimony anyway?
To the broader point, I'm not sure that there are iron rules that an expert who had an opinion about something before can never have that opinion again. In the attorney world, staying on the same side of the "v" should generally avoid any conflicts of interest. But you hope experts start from a tabula rasa. Or at least that they're perceived that way to help their credibility.
I also wonder on the timeline. Case was filed at the end of 2020. The FTC didn't engage Hemphill until 2022. Really? You would think an economist would be helping them from the beginning on the complaint, figuring out which evidence to seek, and so on. So I wonder if what happened -- and this is pure speculation on my part -- was that the FTC DID have such a expert from the beginning, but that expert wasn't going to opine the way they wanted. So they hired Hemphill because his views on this were already known.
But look, all of this might not be that much of a credibility hit. But why even let it be an issue? Why not hire literally anyone else? Those questions make me think Hemphill wasn't actually their first call.
"What was both unsurprising and yet odd about this case was the degree to which it was fought over point number two, with minimal acknowledgement of point number three. That is, it seems clear to me that AT&T made this acquisition with an eye on point number three, yet the government’s case was predicated on point number two; to that end, the government, in my eyes, rightly lost given the case they made. Whether they should have lost a better case is another question entirely."
Carl Shapiro was the FTC's economist expert against Qualcomm. Dan Matheson was at the FTC and I was at Cravath representing Qualcomm. The district court's decision against Qualcomm did not cite Professor Shapiro's testimony once. Then a panel of the Ninth Circuit unanimously reversed.
Cravath was also counsel for Time Warner in the AT&T merger hearing but I did not work on that. Who was AT&T's expert economist there? Why, Dennis Carlton, of course, one of Meta's economists here. Another Meta economist is Catherine Tucker, who was Google's expert on market definition in the Play Store litigation I worked on.
I was thinking the same thing. Did Hemphill prejudge Meta or judge them based on the reporting and evidence he was privy to at the time? Even if the former, that might undercut his credibility as unbiased, but it doesn't undercut the fact that he seemed to be right in his analysis. Hansen's cross is definitely effective in knife-ing Hemphill, but I'm not sure it did much to defend Meta and their conduct.
Expert witnesses are paid and brought by each side. Why do we care if the FTC expert hates Meta or not. I just care about whether his arguments are correct or not. I dont care about his motivation. Its not about believing him. He is not creating facts. He is arguing for the FTC. We know that. I judge his arguments only. I hope the Judge takes the same view.
Hemphill's arguments are what matter to me.
I feel Facebook was crying wolf when it attacked Hemphill's motivation. It just doesnt matter.
We are lucky to have you here and Mr. Stoller did a great job bringing you here.
Now a point from me.
Viewing ads are a cognitive tax -- a mental tax.
Especially on apps like Facebook, Instagram, Snapchat where you are scrolling downward and evaluating every post.
You have to evaluate the post.
That takes mental clock cycles.
Your brain has to decide if it wants to skip over this ad or experience it.
Thats a high cost; it's tiring.
I am surprised and shocked how the defense has gotten away with stating that you can just scroll by it in a second.
Its not that simple.
Your brain processes every post.
Its painful to come across an ad in your feed.
Facebook tells advertisers to disguise their ads so that users cant process they are ads and scroll by quickly.
Facebook tells advertisers do not put text on your ads -- limit the text -- so users cant recognize its an ad and scroll by fast.
Facebook made and makes advertisers run a tool that checks the text on ads to ensure users cant skip them easily.
Facebook is making it harder for your brain to process the ads and skip them.
I hate it when I am scrolling and suddenly what I thought was a post from a friend really is a disguised ad.
Thats a tax.
It stops me.
Its a pain.
I am not hearing this argument loud enough at trial.
Why is Meta/Facebook getting getting away with pretending ads are simple to bypass and dont hurt the consumer?
Its bull.
Its tiring to avoid ads.
Has Judge Boasberg put Facebook and Instagram on his phone and tried using it for a few weeks?
If not how can he ever find out what is going on here?
Seeing 1 add for every 15 posts is way different than seeing 1 for every 4-5 posts.
The good Judge wont ever know that if he doesnt start using Facebook and Instagram.
The FTC is losing a huge opportunity here by not getting the Judge on these apps with his friends and family and really driving home the cognitive mental task involved in skipping advertising.
A second or 2 or 3 every few posts is a hell of a lot of brain power Meta is stealing from us by breaking antitrust laws.
The Judge and the Court and the FTC need to feel the pain to litigate this case properly.
Thanks Franca for being a loyal reader and for your many thought-provoking comments.
There was in fact testimony about this from the current Head of Facebook, Steve Alison, who said that you could spend "all day" scrolling, or scroll through perhaps 10,000 posts, to see all the recent posts from your friends on News Feed. Now that's not because it's ads or friend posts and nothing else, but probably primarily because the algorithm is mostly serving up unconnected content these days or more expanded content from "pages" and "groups". I do think that was helpful for Chief Judge Boasberg to hear that because spending "all day" to see the few pieces of friends/family content you want is a lot different from "just scroll past it" - you need to scroll past it, and keep scrolling, and keep scrolling...
Interesting to see the use of the quantitative SSNIP test from the FTC expert, as we in th EU started to use it a few years back. Is it the first time this kind of test was used? Also i would like to ask about the phrase “qualitative and quantitative evidence and tools” in my countries courts they would apply this phrase that both qualitative and quantitative evidence has to be shown.. so it is different in US law? Another great article and looking forward to more.
Thanks. So just to clear, it was a qualitative HMT analysis. The SSNIP-T or SSNIP-Q (T for terms; Q for quality) was really just looking at whether Meta itself could profitably impose degraded quality, and things like its ranking on consumer sentiments scores or comparing those scores to usage (e.g. after Cambridge Analytica) are potentially ways of looking at that. But as Meta points out, there's no opinion from Hemphill as to what the competitive baseline level of quality, privacy, terms, and the like is. So it's difficult to infer from absolute changes in sentiment whether that's a relative change from the competitive level of sentiment which might be fluctuating for other reasons.
There are definitely other cases where a broader view of the HMT or SSNIP is taken. In Epic Games v. Google (and I represented Epic Games at one point before representing the State Attorneys General against Google in the Play Store case), Epic's expert Doug Bernheim talked about this concept. (Testimony quoted below.)
Bigger picture is that the "neo-Brandeisians" and other advocates of more antitrust enforcement recognize that, because of these problems like a "zero-price" product, two-sidedness, and the Cellophane fallacy, the insistence by the defense side or Chicago school on a quantitative SSNIP is going to result in underdeterrence. There are some cases where the quantitative SSNIP is a lighter lift because there's more data like a merger case or a price-fixing case.
Interesting approach from Hansen in his impeachment of Hemphill. It seems based on how you phrased it, that it was effective in undermining his credibility.
However, it does beg the question, if in fact Hemphill HAD prejudged the outcome of the case, does that in any way reduce the substantive merit of his testimony based on evidence other than his own opinion? Of course it does call into question the validity of Hemphill’s opinion. But if Hemphill’s testimony is primarily his interpretation of factual evidence, isn’t it just as likely that his interpretation is correct or not correct, independent of his views on the case as a whole?
I’ll be interested to see if the judge’s opinion reflects this cross.
I think the issue as Hansen framed it was that if Hemphill had an opinion before the case, he's not doing a neutral evaluation of the evidence and is disregarding evidence that doesn't fit with his preconceived notions. And if all his testimony boils down to is his own interpretation of fact evidence, that gets into the territory of testimony that's not helpful to the court, as experts can't just regurgitate what fact witnesses say or tell the court how it should weigh evidence. Hansen tried to get some answers in this direction, asking Hemphill if he knows better than Meta executives on these topics, is he calling Meta execs liars, etc.
For me, it was effective to see -- even more so in the stuff I found from 2020, rather than the 2019 "roadshow" deck -- that Hemphill was making sweeping conclusions about Facebook with very little to back it up at the time. That's not to say he can't back it up now. But it is telling that despite all that, he doesn't have an opinion on whether Instagram would have done better or worse without being acquired by Facebook. That's kind of the whole point. So if someone who is purportedly biased can't even agree with the FTC's theory on that score, how effective is the testimony anyway?
To the broader point, I'm not sure that there are iron rules that an expert who had an opinion about something before can never have that opinion again. In the attorney world, staying on the same side of the "v" should generally avoid any conflicts of interest. But you hope experts start from a tabula rasa. Or at least that they're perceived that way to help their credibility.
I also wonder on the timeline. Case was filed at the end of 2020. The FTC didn't engage Hemphill until 2022. Really? You would think an economist would be helping them from the beginning on the complaint, figuring out which evidence to seek, and so on. So I wonder if what happened -- and this is pure speculation on my part -- was that the FTC DID have such a expert from the beginning, but that expert wasn't going to opine the way they wanted. So they hired Hemphill because his views on this were already known.
But look, all of this might not be that much of a credibility hit. But why even let it be an issue? Why not hire literally anyone else? Those questions make me think Hemphill wasn't actually their first call.
From what I understand/remember, the FTC had hired Carl Shapiro to be the expert witness, but he left the case once Lina Khan took over the FTC.
https://www.politico.com/news/2021/07/30/facebook-antitrust-suit-carl-shapiro-501825
I had no idea! Thanks for sharing this.
Lina Khan cleaned house and brought honesty to antitrust enforcement.
Carl Shapiro and Makan Delrahim's views on antitrust are horribly out of date.
Under them the DOJ ignored key arguments.
Anyone interested read Ben Thompson on the case:
https://stratechery.com/2018/the-need-for-neutrality/
"What was both unsurprising and yet odd about this case was the degree to which it was fought over point number two, with minimal acknowledgement of point number three. That is, it seems clear to me that AT&T made this acquisition with an eye on point number three, yet the government’s case was predicated on point number two; to that end, the government, in my eyes, rightly lost given the case they made. Whether they should have lost a better case is another question entirely."
The Judge really took it to Shapiro:
https://variety.com/2018/politics/news/att-time-warner-carl-shapiro-merger-predictions-1202750523/
I remember the case well.
DOJ lost
AT&T won the right to buy Time Warner.
Ironically AT&T got hurt by overpaying for Time Warner:
AT&T Dumps Time Warner Business Four Years After $85B Deal
Carl Shapiro was the FTC's economist expert against Qualcomm. Dan Matheson was at the FTC and I was at Cravath representing Qualcomm. The district court's decision against Qualcomm did not cite Professor Shapiro's testimony once. Then a panel of the Ninth Circuit unanimously reversed.
Cravath was also counsel for Time Warner in the AT&T merger hearing but I did not work on that. Who was AT&T's expert economist there? Why, Dennis Carlton, of course, one of Meta's economists here. Another Meta economist is Catherine Tucker, who was Google's expert on market definition in the Play Store litigation I worked on.
I was thinking the same thing. Did Hemphill prejudge Meta or judge them based on the reporting and evidence he was privy to at the time? Even if the former, that might undercut his credibility as unbiased, but it doesn't undercut the fact that he seemed to be right in his analysis. Hansen's cross is definitely effective in knife-ing Hemphill, but I'm not sure it did much to defend Meta and their conduct.
Expert witnesses are paid and brought by each side. Why do we care if the FTC expert hates Meta or not. I just care about whether his arguments are correct or not. I dont care about his motivation. Its not about believing him. He is not creating facts. He is arguing for the FTC. We know that. I judge his arguments only. I hope the Judge takes the same view.
Hemphill's arguments are what matter to me.
I feel Facebook was crying wolf when it attacked Hemphill's motivation. It just doesnt matter.
Thank you Mr. Benedict for your writing.
I look forward to your analysis and wait for it.
We are lucky to have you here and Mr. Stoller did a great job bringing you here.
Now a point from me.
Viewing ads are a cognitive tax -- a mental tax.
Especially on apps like Facebook, Instagram, Snapchat where you are scrolling downward and evaluating every post.
You have to evaluate the post.
That takes mental clock cycles.
Your brain has to decide if it wants to skip over this ad or experience it.
Thats a high cost; it's tiring.
I am surprised and shocked how the defense has gotten away with stating that you can just scroll by it in a second.
Its not that simple.
Your brain processes every post.
Its painful to come across an ad in your feed.
Facebook tells advertisers to disguise their ads so that users cant process they are ads and scroll by quickly.
Facebook tells advertisers do not put text on your ads -- limit the text -- so users cant recognize its an ad and scroll by fast.
Facebook made and makes advertisers run a tool that checks the text on ads to ensure users cant skip them easily.
Facebook is making it harder for your brain to process the ads and skip them.
I hate it when I am scrolling and suddenly what I thought was a post from a friend really is a disguised ad.
Thats a tax.
It stops me.
Its a pain.
I am not hearing this argument loud enough at trial.
Why is Meta/Facebook getting getting away with pretending ads are simple to bypass and dont hurt the consumer?
Its bull.
Its tiring to avoid ads.
Has Judge Boasberg put Facebook and Instagram on his phone and tried using it for a few weeks?
If not how can he ever find out what is going on here?
Seeing 1 add for every 15 posts is way different than seeing 1 for every 4-5 posts.
The good Judge wont ever know that if he doesnt start using Facebook and Instagram.
The FTC is losing a huge opportunity here by not getting the Judge on these apps with his friends and family and really driving home the cognitive mental task involved in skipping advertising.
A second or 2 or 3 every few posts is a hell of a lot of brain power Meta is stealing from us by breaking antitrust laws.
The Judge and the Court and the FTC need to feel the pain to litigate this case properly.
Thanks Franca for being a loyal reader and for your many thought-provoking comments.
There was in fact testimony about this from the current Head of Facebook, Steve Alison, who said that you could spend "all day" scrolling, or scroll through perhaps 10,000 posts, to see all the recent posts from your friends on News Feed. Now that's not because it's ads or friend posts and nothing else, but probably primarily because the algorithm is mostly serving up unconnected content these days or more expanded content from "pages" and "groups". I do think that was helpful for Chief Judge Boasberg to hear that because spending "all day" to see the few pieces of friends/family content you want is a lot different from "just scroll past it" - you need to scroll past it, and keep scrolling, and keep scrolling...
I valued Alison's testimony.
Scrolling thru 10000 posts is holding your friends and family hostage while forcing you to review all the other posts they throw at you.
I also think that Facebook hiding that ads in your feed are ads on purpose is a very important point.
It harm to mentally decide whether each post is an ad that you want to skip.
Cognitive taxes from Meta.
The FTC must show harm.
I hope the FTC is being aggressive enough about the harms so Judge Boasberg feels it.
Personally, I found the ad load increase from .3% to 20+%, in a relatively short time frame, to be quite shocking.
All in all, thus far, as a passive observer, the government seems to have proven its case, but not without missteps and speed bumps.
I still don't have a strong view one way or another as to who's ahead. But I do think the FTC is doing a lot better than the conventional wisdom says.
Interesting to see the use of the quantitative SSNIP test from the FTC expert, as we in th EU started to use it a few years back. Is it the first time this kind of test was used? Also i would like to ask about the phrase “qualitative and quantitative evidence and tools” in my countries courts they would apply this phrase that both qualitative and quantitative evidence has to be shown.. so it is different in US law? Another great article and looking forward to more.
Thanks. So just to clear, it was a qualitative HMT analysis. The SSNIP-T or SSNIP-Q (T for terms; Q for quality) was really just looking at whether Meta itself could profitably impose degraded quality, and things like its ranking on consumer sentiments scores or comparing those scores to usage (e.g. after Cambridge Analytica) are potentially ways of looking at that. But as Meta points out, there's no opinion from Hemphill as to what the competitive baseline level of quality, privacy, terms, and the like is. So it's difficult to infer from absolute changes in sentiment whether that's a relative change from the competitive level of sentiment which might be fluctuating for other reasons.
There are definitely other cases where a broader view of the HMT or SSNIP is taken. In Epic Games v. Google (and I represented Epic Games at one point before representing the State Attorneys General against Google in the Play Store case), Epic's expert Doug Bernheim talked about this concept. (Testimony quoted below.)
Bigger picture is that the "neo-Brandeisians" and other advocates of more antitrust enforcement recognize that, because of these problems like a "zero-price" product, two-sidedness, and the Cellophane fallacy, the insistence by the defense side or Chicago school on a quantitative SSNIP is going to result in underdeterrence. There are some cases where the quantitative SSNIP is a lighter lift because there's more data like a merger case or a price-fixing case.
Thanks
Here is the Bernheim testimony from the Epic Games v. Google trial I was thinking about in response to your question:
Q. Exactly. And so the SSNIP test isn't measuring what
consumers would do if the quality of the Google Play Store went
down; right?
A. If you're doing it in price, you shouldn't be doing it in
quality at the same time, although in principle you can also do
a SSNIP test varying quality instead of price.
Q. Right. You didn't do that, though?
A. I think I wrote something about that in one of the
reports. I don't think it really makes a difference because
quality can be translated into monetary terms, which means that
you can kind of treat it as a price. You reach the same
conclusions.
Q. But at a higher level, an economist doesn't even need to
use the SSNIP test to define a relevant market; right?
A. Sometimes economists define markets entirely based on more
qualitative approaches, evaluating substitution, but the
SSNIP -- the SSNIP idea remains the lens that we think about
when we consider the magnitude of substitution.
Q. And, in fact, you, yourself, have testified about the
relevant market in other cases without doing the SSNIP test
that you described to the jury today; right?
A. I think you're referring in particular to the hypothetical
monopolist test, which involves a little bit more than a SSNIP.
And, yes, I've testified in other cases without doing a
hypothetical -- a quantitative hypothetical monopolist test.
Thanks for your great work. I love how active you are even in the comments.