“Poor Mr. Zebley:” Both Xitter’s Lawyers and Journalists Responding to Boilerplate Need to re-Read Mueller
I’ve stopped trying to convince Russian denialists on Xitter that they’re willfully ignorant of facts. At this point, denialists are just trolls exploiting Xitter’s algorithm to create scandal.
I try to focus my time, instead, on conspiracy theorists platformed by prominent schools of journalism.
But when others try to correct denialists on Xitter, they almost always say the denialists haven’t read the Mueller Report closely enough.
So I found it wildly ironic that Chief Judge Beryl Howell, during a period in February when Elon Musk was letting denialists like Matty Dick Pics Taibbi invade the privacy of then-Twitter’s users so he could spew conspiracy theories, Howell scolded Twitter’s lawyer George Varghese that he hadn’t read the Mueller Report closely enough.
THE COURT: You need to read the Mueller report a little bit more carefully.
The transcript of the court hearing and much of the rest of the back-up to Xitter’s attempt to stall compliance of the warrant was unsealed yesterday.
Mind you, Howell was trying to convey to Twitter’s team that there is precedent for investigating Donald J. Trump without giving him advance warning of every investigative step.
MR. VARGHESE: Yes, Your Honor. Our —- by
THE COURT: You think that for 230 orders, 2800 subpoenas, and 500 search and seizure warrants the Mueller team gave advance notice to the former President of what they were about?
MR. VARGHESE: I don’t know that, Your Honor.
THE COURT: You do not know that.
The hearing made it pretty clear that Howell is convinced that Trump will stop at nothing to obstruct criminal investigations into himself.
Howell, who knows what went into the Mueller Report as well as anyone outside the investigative team, does know that.
In fact, when she told Varghese he should have read the Mueller Report more closely, she had just pointed to private comms described in the Mueller Report — the ones where Trump told Mike Flynn to stay strong — where Trump had not gotten advance notice, as prosecutors were demanding he not get advance notice about a warrant to Twitter.
THE COURT: Because the Mueller report talks about the hundreds of Stored Communications Act — let me quote.
Let’s see.
The Mueller report states that: As part of its investigation, they issued more than 2800 subpoenas under the auspices of the grand jury in the District of Columbia.
They executed nearly 500 search and seizure warrants, obtained more than 230 orders for communications records under 18 U.S.C. Section 2703(d); and then it goes on and on and on for all of the other things they did.
And some of those communications included the former President’s private and public messages to General Flynn, encouraging him to “Stay strong,” and conveying that the President still cared about him, before he began to cooperate with the government.
So what makes Twitter think that, before the government obtained and reviewed those Trump-Flynn communications, the government provided prior notice to the former President so that he can assert executive privilege?
MR. VARGHESE: My understanding, Your Honor, is that the Mueller investigators were in contact with the White House counsel’s office about executive privilege concerns.
THE COURT: You quoted the one part that said that, and that was for testimony, testimony, where it was not covert.
Side note: Xitter’s lawyers may not have been entirely wrong about consultations with the White House counsel, even for materials obtained covertly.
This exchange happened on February 7. Two days later there was a follow-up hearing, and WilmerHale counsel Aaron Zebley — someone who knows better than Beryl Howell what happened to the materials for which Howell approved legal process after it got handed over but before they ended up in the Report itself — filed an appearance in this challenge. He never spoke though; he showed up late, if at all, and at one point, after Twitter had presented their opening argument, Howell asked someone to check whether “poor Mr. Zebley” was standing outside a locked door waiting to get in.
THE COURT: Okay. Well, let me just —
Mr. Windom, do you want to think about that or do you want to respond?
Do you think Mr. Zebley is standing outside the locked door?
MR. HOLTZBLATT: I think there is a chance.
THE COURT: Could you check? Poor Mr. Zebley.
MR. WINDOM: Should I wait, Your Honor, or proceed?
THE COURT: Proceed. In my chambers we wait for no man.
Twitter was trying to make an argument that someone had to attend to potential Executive Privilege claims. Howell and the prosecutors nodded several times to a filter protocol addressing privilege issues, of which Twitter was ignorant. And yet Twitter was refusing to comply unless they had the opportunity to tell Donald Trump about the warrant in advance.
Beryl Howell, who was years into her second investigation of Donald Trump at this point, might be forgiven for impatience with lawyers who don’t understand how many Executive Privilege disputes she had presided over between those two investigations. They might be forgiven for their ignorance of all the resolutions of Trump’s current challenges to Executive Privilege in the January 6 investigation.
That said, Twitter’s lawyers aren’t the only ones who should have read the Mueller Report more closely. So are the journalists reporting on this.
One after another journalist (CNN, NYT, Politico, all involving journalists who covered the Mueller investigation) has mistaken DOJ’s request for data — attachment B to the warrant — as some kind of statement of what DOJ was most interested in receiving. Based on that, their stories focus on the fact that DOJ asked for or obtained DMs involving the former President.
But that attachment looks to be largely boilerplate. It is not much different from warrants obtained five years ago, in the Mueller investigation, such as this one, also served on Twitter, apparently targeting Trump’s rat-fucker Roger Stone in an investigation into whether he was serving as a foreign agent of Russia, a warrant that also came with a gag, one Twitter did not contest. One main — telling — difference, is that the Trump request included standard subscription information, which Mueller’s investigators appear to have already requested; one of the items on which Twitter held up compliance, in fact, was Trump’s gender, a sure testament to obstruction within the company.
While Twitter’s services have changed significantly in the interim years, both ask for the same kind of information: DMs, drafts, deleted content, favorited content.
And for good reason!!! These warrants may well have been targeting the same kind of behavior, the kind of organized troll campaigns that exploit Twitter’s algorithms, in which users use a variety of means to obscure their identity. There is a significant likelihood these warrants were targeting precisely the same group of far right online activity, the very same people.
One of the most important Twitter users leading up to January 6, Ali Alexander, is the protégé of Roger Stone and the effort to drive attendance at January 6, Stop the Steal, was a continuation of the effort Stone started in 2016, an effort that may well have been covered by that 2018 warrant or one of the others targeting Stone’s Twitter activity.
To be sure: There are DMs in Trump’s account, though it’s not entirely clear when they date to. Without reading any of the DMs, Twitter checked to see whether the volume of data in Trump’s account indicated the presence of DMs.
MR. VARGHESE: So, Your Honor, we went back — because this was an important issue for us to compare, whether or not there were potentially confidential communications in the account, and we were able to confirm that.
THE COURT: How?
MR. VARGHESE: So, Your Honor, there was a way that we compared the size of what a storage would be for DMs empty versus the size of storage if there were DMs in the account. And we were able to determine that there was some volume in that for this account. So there are confidential communications. We don’t know the context of it, we don’t know —
THE COURT: They are direct messages. What makes you think — do you think that everything that a President says, which is generically a presidential communication, is subject to the presidential communications privilege?
MR. VARGHESE: No, Your Honor.
But Twitter’s focus on DMs arose from their frivolous basis for delaying response to the warrant — their claim that some of these DMs might be subject to a claim of Executive Privilege.
Moreover, having DMs in the account is not the same thing as a prosecutor confirming that they ultimately obtained DMs, or that any DMs were relevant to the investigation, or that DMs were one of the things they were most interested in.
I don’t doubt that’s likely! But what prosecutors asked for and what was in the larger account is not the same thing as what DOJ ultimately received and used.
And the DMs — most of them, anyway — are something that were available elsewhere. At least as represented in the dispute, NARA already has Trump’s DMs from the period (DOJ chose not to go to NARA, in part, because they wanted to avoid notice that NARA has provided to Trump along the way).
There were three more things that DOJ showed perhaps more interest in, requiring Twitter to go beyond their normal warrant response tools to comply.
The first has to do with emails to Twitter about the account, of which prosecutor Thomas Windom was most interested in emails from people on behalf of Trump.
But this information about, you know, what it is that we say that we’re most specifically interested in, I did not represent that we were most interested in communications betueen government officials and Twitter regarding the account.
We did point out that — much as Your Honor did just now — it seemed beyond comprehension that there weren’t communications regarding the account when it was suspended and terminated, but that doesn’t mean government officials at least cabined to that. It can mean campaign officials. It can be anybody acting on behalf of the user of the account, or the user of the account himself.
THE COURT: So any person regarding the account is broader than what you just said, though, Mr. Windom.
“Any person regarding the account” is quite broad. It could be all the complaints of all of the Trump supporters out in the world saying: What are you doing, Twitter?
So I take it, from what you just said, that you are interested only in =- rather than “any person,” a person who was the subscriber or user of the account or on behalf of that person regarding the account?
MR. WINDOM: Yes, ma’am. An agent thereof.
When Twitter cut Trump off in 2021, they cut off active plans for follow-up attacks. And these emails might indicate awareness of how Trump was using Twitter as a tool to foment insurrection.
Another item on which Windom focused in the following hearing was associated accounts — other accounts the identifiers used with Trump’s accounts also use. Twitter claims they don’t have that — at least not in their law enforcement portal — and so had to collect it manually. But DOJ did ask them to produce it. (Note, the fact that Xitter doesn’t store this is one reason why they’re so bad at tracking information operation campaigns, because visibility on these kind of associations are how you discover them.)
MR. HOLTZBLATT: Well, Your Honor, we don’t — the issue, Your Honor — there isn’t a category of “associated account information”; that’s not information that Twitter stores.
What we are doing right now is manually attempting to ascertain links between accounts. But the ascertainment of links between accounts on the basis of machine, cookie, IP address, email address, or other account or device identifier is not information that Twitter possesses, it would be information that Twitter needs to create. So that’s the reason why we had not previously produced it because it’s not a category of information that we actually possess.
[snip]
MR. WINDOM: It is, as explained more fully in the warrant — but for these purposes, it is a useful tool in identifying what other accounts are being used by the same user or by the same device that has access to the account is oftentimes in any number of cases, user attribution is important. And if there are other accounts that a user is using, that is very important to the government’s investigation.
[snip]
MR. HOLTZBLATT: That’s right. If the records — if the linkage between accounts, which is what we understand this category to be referring to, is not itself a piece of information that we keep, then it’s not a business record that we would ordinarily produce.
What I understand the government to be asking is for us to analyze our data, as opposed to produce existing data. And we are trying to work with the government in that respect, but that is the reason that it is not something that — that is a different category of information.
As Windom explained, this information is critical to any attribution, but it’s also important to learning the network of people who would Tweet on Trump’s behalf, and any overlap between his account and their own (as Roger Stone’s showed in 2016).
Then there’s something that remains only partially explained. For some reason — even Twitter could not figure out why — there were two preservations of Trump’s account in January 2021, before the preservation associated with this warrant. One was on January 9. The other covered January 11 and 12. And when asked, the government of course wanted the latter preservation too — and it is in the possession of Twitter, and so covered by the warrant.
MR. HOLTZBLATT: At 5 p.m. on February 7th, I think that was our day, we produced all data in this category that was in the standard production tools of Twitter.
We communicated with the government on February 8th that there were prior preservations of the subject account that are not within Twitter’s standard production tools and that would, therefore, require engineering to obtain information. And we asked the government whether it wished us to undertake that effort, and the government confirmed that it did.
And we have since then — when we produced on February 7, we indicated to the government in our production letter that there was potentially deleted data that might exist, which is what would be found in prior preservations, but that it would require additional engineering efforts.
At 2 a.m. last night, or this morning, Twitter produced additional information from those prior preservations that falls within category 2A. There are —
THE COURT: When you say “prior preservations” what are you talking about?
Prior litigation holds of some kind or that you had a stash or a cache of preserved data sitting in different places? What are you talking about?
MR. HOLTZBLATT: I am referring — with respect to this particular account, I am referring to preservations from two specific dates. There is a preservation that was made that includes the subject account covering January 3rd to 9th, 2021. There is a second preservation of this that includes this account that covers January 11 to 12, 2021.
Those are collections of data that — they are not — it’s not coterminous with the categories that would exist in the active account right now and — and that’s data that does not exist within a production environment. So it’s not data that you can just click — we have a system to just click a button and produce, which is why we indicated that further engineering efforts might be necessary.
We asked the government if they wished us to undertake those efforts. We had an engineer working through the night, after the government asked us to, to undertake those efforts. At 2 a.m. in the morning we produced additional information that came from those preservation.
There are two categories of information that — actually, I’m sorry, three categories of information that we are still working to produce because of the engineering challenges associated.
One of those categories is the list of — I am not sure this is from 2A. But I think, for purposes of coherence, it would be helpful for me to describe it now because it connects to this preservation; that is, followers — a list of followers for this account that were contained within the January 11 through 12th prior preservation. We have segregated that information. It is a complicated and large set of information. And we are unable to deliver it in the manner that we normally deliver information to law enforcement, which is to send a token.
We believe right now it would require physical media to put that information on and to hand it over to the government.
[snip]
MR. HOLTZBLATT: As I mentioned, Your Honor, there were two prior preservations, and then there is the current production tools. In two of the three of those sets, the January 3 through 9 and the current one, we have produced the tweets and related tweet information for the account.
In the January 11 to 12th prior preservation, the way that the tweet and tweet-related information is stored, it goes all the way back to 2006. We don’t have a warrant — that is contents of user communications. He don’t nave a warrant that would permit us to produce the entirety of that information. So what we have is a tool 7 that — what we refer to as a redaction (sic) tool or trimming tool. Because this is not a production environment, a human being has to go in and manually trim the information to isolate the date range. That, I think Your Honor can understand, is a laborious process, including for this particular account, given the time frame; and we need to isolate it, I think, over a three-month, four-month period, I’m sorry, Your Honor. So we are undertaking it.
Unsurprisingly, DOJ wanted to be able to compare the accounts as they existed on January 8 and January 12, 2021, because Trump’s attack was still ongoing and because people were beginning to delete data.
Trump’s DMs, if he used them or even just received them in this period, would be critically important. But Twitter was one of Trump’s most important tools in sowing an insurrection. And the data showing how he used the account, and who also used it, is as important to understanding how the tool worked as the non-public content.
The transcript reads like a scolding of a child. Howell instructs the Xitterans to “Control F” through their data at one point and also asks them if they delay in producing similar data for other users. Of course, they can’t answer that.
Did I also read correctly that the Xidiots were asked to produce any and all communications between they and anyone regarding the account? Is that part of the initial request or to uncover if there were attempts to warn someone about the DOJ review of the data?
[Welcome back to emptywheel. Please use the same username and email address each time you comment so that community members get to know you. “Wilmington Frisbee Guy” is your third user name; you last commented as “WilmingtonFrisbeeGuy.” The spaces make a difference. Thanks. /~Rayne]
Buttinski here. I don’t understand why so many emptywheel users have this problem. I use a standard browser, Firefox, that remembers past logins to websites and offers these as prompts for the username ans password. Are your hamfisted readers disabling this useful function?
It is no longer practicable to remember secure credentials for all the sites one visits. The non-exclusive choices are : to write everything down on a dog-eared sheet of paper: to use insecure credentials, recycling a very few names and passwords; to let the browser remember for you. FWIW, not much to an expert like Marcy, I limit my high-security credentials to a small number of key sites – mainly banks, intermediaries like Google and Facebook, the server that hosts a personal website. What’s the real downside to a third party hacking my identity here? To give an idea of the cognitive baseline here, I can’t remember the phone numbers of my wife and three children, though of course the phone does.
First, they’re not logins. If people flush their cache memory, change browsers or devices, they may not have the same autofill data available.
Second, because they’re not logins they’re not likely to be kept in password manager apps, but because they’re not logins, keeping a username/email address on a plain paper notepad is not a security risk to this site.
Third, the possibility someone could spoof a user’s identity is a realistic threat when users rely on extremely common names (like “James”) and then troll in comments before they can be screened out. It’s a genuine concern, one I’m not going to elaborate on for you. No one is forcing you or anyone else struggling with username consistency to comment here.
As for your cognitive baseline and your family’s phone numbers…*smh* So much for emergency preparedness.
tiny typo in 2nd sentence of last paragraph: should be: one of (one fo)–fast typist you are Dr Wheeler. Thanks for all of this. all of it.
Typo? Okay, fair enough. Now, how do you pronounce “Xitter”? Like “Xi” in “Xi Jingping”? That would be “Zitter,” which seems to fit as a synecdoche, I suppose
Xi => Shee
Jinping = J as In John
=>
Jeen-peeng
Xitter => e after t=> uh
=> Shee-tuhr
Seriously? You must be xitting me.
Zhèngquè
https://pages.ucsd.edu/~dkjordan/resources/PronouncingMandarin.html
Reading the transcript of the hearing in Howell’s courtroom makes me long for video and sound. The tone of voice and the look on her face as she says “You think that for 230 orders, 2800 subpoenas, and 500 search and seizure warrants the Mueller team gave advance notice to the former President of what they were about?” is likely priceless.
Honestly, I would always laugh when calling into hearings before her. Her tone is a cross of school marm and Top 500 CEO, with zero time for nonsense.
So yeah, I kept hearing her voice as I read this.
“In my chambers we wait for no man.”
I had more than one teacher who ran their class like this. If you wanted a delay or extension on an assignment because your grandmother died, they wanted to see a clipping of your grandmother’s obituary.
From a teaching colleague, many years ago: “If you think English courses are tough on students, they’re murder on Grandmas.”
“In my chambers we wait for no man.”
When I read that, after a loud chuckle, I thought, “That’s t-shirt worthy!”
Thank you for this post EW. Eye opening!
I thought of Lord of the Rings: Return of the King when Éowyn confronts the Witch-King.
Might have to give Judge Howell the nickname Éowyn. LOL
One of my fave moments in the books: the power of folk-tale motifs elevated to an Aristotelean reversal. Loved it; got tingles the first time I read it.
I loved her comment “this Court waits for no man.” Priceless.
I wish I could hear her voice. I’ve read her rulings, so I’m familiar with her writing voice in those pieces. But our voices are such unique evocations of our entire selves: our bodies, our emotions, our intentions. Nothing compares to being able to hear a human’s speaking voice.
Your wish is my command!
Starting around 36:30 minutes in:
https://youtu.be/1Sagz-K0AXA
I had to read that twice to stop my brain from changing it to “cross school marm” which my Mom was.
These vivid transcript excerpts are great illustrations of the iron-fisted style in which most federal district court judges preside over their courtrooms and everyone present. It’s not for the weak!
Truly! To say her comments were withering is an understatement, i.e., when Judge Howell, referring to Attachment B, tells Holtzblatt, “We’re just going to go through it line by line….It seems like that kind of supervision of Twitter is needed here.” Or, a bit later when she says, “Do I have to read it to you or can you read it yourself?” Ouch!
Second time:
The Court: Do you think Mr. Zebley is standing outside the locked door?”
Mr. Holtzblatt: I think there is a chance.
The Court: Could you check? Poor Mr Zebley.
Mr. Windom: Should I wait, Your Honor, or proceed?”
The Court: Proceed. In my chambers we wait for no man.
Judge Howell to Mr. Holtzblatt:
“Is there a reason why Ms. Russell just has to sit there as opposed to speaking since she seems to be the person with the answers?”
The Xitter lawyer isn’t exactly telling the truth about associated accounts in my experience. I used to maintain a pile of accounts, serious, scientific, personal and troll accounts. Specifically I had account that I used only to post photographs of Max Headroom any time Joe Walsh posted something insanely fascistic which was frequently. Apparently I pissed him off and I received an IP ban essentially, all of my accounts, accessed from different machines typically (work from work right), all of them were banned AND all of my private domains, which I’d used to set up the accounts were banned from creating new accounts. Any email address that uses @hypperbone.com was unable to complete registration of an account, this was still true a few years ago as well. So xitter certainly CAN parse associations if it wants to become it specifically did that to me.
Also, Joe Walsh still looks like a facist version of Max Headroom.
I know enough about database design and management to be dangerous, if not useful, and the story from that Xitter lawyer, even allowing for the inevitable distortions into layspeak, sounds hinky as hell. They’ve got data dumps related to Trump and no one there knows why they have them? And their argument is parsing them is hard? Boo hoo.
I agree. I’ve seen other cases where they clearly knew what accounts were associated. And their work on Roger Stone and the Proud Boys would have required such associative accounts.
It’s possible that Elmo deprecated that.
It has never been clear to me why he took over Twitter and destroyed it on purpose. Rich people usually don’t like to lose wealth in spectacular public ways. The only upside I can see for Xitter’s destruction is quid pro quo.
I would post a picture of the Stay-Puft Marshmallow Man to Chris Cillizza’s tweets. I don’t know that he ever noticed, though, much less got what I was saying.
Exactly. It’s well known that if you use the same email address, device and sometimes even IP address that all your accounts will get banned if one gets banned. And any attempts to make a new account from the same email, device or sometimes even IP will fail or be immediately banned. That Xshitter is claiming they don’t track this info in any way is laughable.
[Welcome back to emptywheel. SECOND REQUEST: Please choose and use a unique username with a minimum of 8 letters. We are moving to a new minimum standard to support community security. You used your email as your username on this comment which may have been unintended (and is not a recommended practice). It has been temporarily changed to match that of your first know comment until you have a new compliant username. Thanks. /~Rayne]
His guitar work on “Hotel California” is still outstanding, though.
;-)
As far as implementation on Twitter’s end I do not know. But as far as what is, by industry (computer science/language/data structures) standards, doable/possible with plenty of case studies, you are 100% correct.
I’m 15 years removed from doing this work, no longer expert. But the strategies I’m sure remain, at worst, similar. With that said, I’ll try and explain briefly the mechanics of how collecting data DOJ wanted and X (at least partially) obfuscated (HOLTZBLATT’s statements) is done.
2 things to know:
a) a little about SQL (RDBMS… eg. Oracle, IBM DB2, MySQL, Postgress (open source)) databases.
b) computer programmed data structures, in this case a category called: TREE(s).
WRT a): it is a simple matter to write recursive SQL commands to search a database for the information requested. Recursive means:
1) search
2) search again for each result from prior search, saving each result in a TREE node (see below) after each search.
3) repeat (called a loop) until results return empty (eg. nothing).
So search criteria simply what DOJ asked for.
The TREE structure is more important to get how this model (find DOJ requested data) works. I found a pretty decent write up of how this works here. Reading through 1st section: What is Tree Data Structure?, should convey the idea to most folks.
A word (from TREE link) about: NODE(s). If you look at the top chart in the link, each circle is a “NODE”. For my purpose here (there are exceptions), each node is designed to contain basic information for navigating the TREE. Each node will “know” all the nodes it is directly connected to. The computer code uses this information to navigate around the TREE. The data being searched for can be stored in a NODE as well, but usually
There are a lot of different tree structures. We had 2 good size volumes on this alone in our shop. Some structures can do multiple things fairly well, but when dealing with massive amount of data something like this would produce a good programmer would know structure best suited for the task at hand.
Most of what you say is very true, but if there’s one thing SQL is not good for, it’s recursion. You’d most likely use some other language tool to interface with the database and do recursive calls.
Thx. I was not trying to write a manual, just quick overview so people could get the general idea. E.G. that it is not, technically difficult to produce the kinds of results DOJ was asking for. I imagine as frequently as they (at least pre-Elmo) have to produce stuff they have all this done behind a nifty front end (UI) that lets a user just enter search parameters.
Decent short writeup on recursive SQL, also called: SQL CTE… here: https://medium.com/swlh/recursion-in-sql-explained-graphically-679f6a0f143b. We did that a lot, works well.
Great article. Early in my reading of this piece I could not help but wonder when the unredacted Mueller report might be made public?
Everything you’d want to read, except the referrals, as been made public, largely thanks to the tireless FOIA work of Jason Leopold.
For those of us who read the original report the day it was released, how important is it to reread with FOIA-liberated additions? I’m asking anyone who’s kept up, not just EW.
Here’s the most recently released full report.
https://www.documentcloud.org/documents/20401920-201102-mueller-report
There were 4 pages released after that showed that Mueller issued declinations for Don Jr (misdemeanor CFAA) and JD Gordon (foreign agent).
I’ve done posts on the interview reports of Gates, Bannon, Flynn, and others.
A few things stood out to me. First, Xitter actually made an engineer wake up at 2:00am and work through the night simply because they slow-walked a response and then had to scramble? That is toxic beyond belief! Why would they treat people this way?
Second, speaking for how our Large Company does it, we use a third party observability platform called Splunk. With Splunk, you log events such as account activity, API calls, and any other request span you think will help in understanding a live service under production traffic. This is called “distributed tracing” and it’s how something like Xitter, with its “crazy stack” of “microservices” would handle monitoring the system.
Several times per day, every day, we do Splunk queries to investigate issues. Once you write your query, the results can be exported as a table or even converted into a dashboard for easy repetition of tasks if the problem comes up again. I sincerely question the idea that although this wasn’t a query they’d already written and saved to a dashboard, it wasn’t equally trivial for an engineer to query this data and dump the results and that this required some immense delay or an emergency page to a sleeping employee.
Xitter sounds like a toxic workplace and their lawyers sound like liars.
“We asked the government if they wished us to undertake those efforts. We had an engineer working through the night, after the government asked us to, to undertake those efforts. At 2 a.m. in the morning we produced additional information that came from those preservation.”
Not that it particularly matters since I agree that the corporate behavior is toxic either way, but the above paragraph to me implies that they required an engineer to stay late after their workday should have ended, working until they produced the information at 2 A.M., not that the engineer was awakened at 2 A.M.
I highly suggest you listen to Behind the Bastards pod about Elon and all the illegal real estate shannangins with his employees living in the Twitter building. He basically trapped them in the building with shoddy electrical and floor heaters, and no way for fire dept to get to them bc nothing has been built to code. That is just one example.
The list goes on and on about how unbelievably horrible his management skills are and his total lack of concern for his employees’ health.
I always knew he was bad, but he is Bangladeshi working codes bad. Leona Helmsley has nothing on him and people still revere this horrible excuse for a human as some sort of genius.
Excellent piece! Thank God for MW and Judge Beryl Howell.
Rayne, I wrote a long-ish reply that got swallowed into moderation. Did I use a keyword I shouldn’t have? Something else I need to avoid in the future?
This is a bit off-topic, & proof of my age & tech-challenged status, but it seems likely to become more pertinent as discovery moves ahead: In the context of criminal discovery, what is a “scoped” search warrant return? The gov’t mentioned “filtered” & “scoped” discovery productions in yesterday’s supplemental response to the standing discovery order in the SDFL prosecution, https://storage.courtlistener.com/recap/gov.uscourts.flsd.648652/gov.uscourts.flsd.648652.122.0.pdf. I inferred that “filtered” meant stuff that had been screened by the filter team doing privilege reviews, & it seemed likely that “scoped” meant that the gov’t production had been limited in some way, maybe to correspond with the scope of the discovery order, but the term was new to this retired fed. All non-anatomical suggestions will be gratefully received!
So Xitter would turn over everything that complied with the October 1 to January 20 timeframe.
Then a filter team would first take anything privileged out of it. It was clear that there was some agreed upon protocol in place tied to the warrant.
Then the investigative team would take everything that wasn’t privileged and pull out what met the scope of the warrant — so stuff that pertained to stealing the election, but not stuff that pertained to bombing Iran (as one hypothetical). Only the stuff within scope would be loaded up into investigative servers.
Many thanks!
How do we know that Elmo didn’t warn Trump? (outside of the fact that Trump can’t keep a secret) and surely would have blurted out something by now).
A fine example of a colleague’s observation that, within the walls of his or her courtroom, there is no more powerful person in the United States than a federal district judge.
It has never been clear to me why he took over Twitter and destroyed it on purpose. Rich people usually don’t like to lose wealth in spectacular public ways. The only upside I can see for Xitter’s destruction is quid pro quo.
Can you not see a reason in this post itself? That by purchasing and destroying the former Twitter — with the help of financing by Saudi and Qatar money — evidence of election interference past and future can be hidden from government and the public?
Have you forgotten, too, that Twitter had been breached from within by Saudi spies who’d been hired by the platform?
Which brings us to the heart of this post: why should the former Twitter give advance notice to Trump what the DOJ has subpoenaed when Trump has likely exposed classified information to hostile foreign entities because he’s a goddamned human sieve — and likely a useful sieve to the Saudis and Qatar as well as Russia?
That is an insult to sieves, Rayne. Sieves have the ability not to leak some material!
If it has not yet been said: poor Judge Howell (and the others overseeing J6-related trials). I wouldn’t blame her a bit for being a bit snappy! I’m grateful for her overseeing the defense of our Constitution in sober fashion without being pushed around. Thank you Marcy for these illuminating comments.
Xitter does in fact track the “linkage between accounts”. If an account gets permanently suspended, one of the ways that Xitter ensures the banned individual does not circumvent the suspension is by monitoring newly created accounts. If a new account is accessed by a client that has also logged into accounts associated with a permanently suspended account, a suspension is placed on the newly created account. This is well known to every troll on the platform.
[Welcome back to emptywheel. Please use the same username and email address each time you comment so that community members get to know you. “Open Source Fact Checker” is inadequately unique, please use a different username which is at least 8 letters in length. Thanks. /~Rayne]
I’m (partly) a data manager by trade, so here’s some context to what the lawyers are talking about.
To start with the end, when Holtzblatt/Twitter says “we normally deliver information to law enforcement(…) is to send a token”, what he means is that Twitter has a law-enforcement portal that when presented with a key/token will give access to the live account information related to said token. I’m assuming that Twitter would still have to do some back-end work to connect which accounts and type of information the token should grant access to, but this would probably also be done using a dedicated web-portal, which is what Holtzblatt/Twitter referred to as “a system to just click a button and produce [the requested data]”.
The previously preserved data would not be a live version of the accounts (which could then be altered), but stored either in a segregated part of the system or more likely extracted into some form of container. Delivering a copy of this to law enforcement in physical form would not be a particularly onerous task. Complaining that “a human being has to go in and manually trim the information to isolate the date range” is being rather disingenious, as it would be a pretty straight-forward data management job to extract a subset. Exactly which tools are fit for the job varies with how the data is stored.
Holtzblatt/Twitter earlier said “the ascertainment of links between accounts on the basis of machine, cookie, IP address, email address, or other account or device identifier is not information that Twitter possesses, it would be information that Twitter needs to create” he’s trying to conflate these analyzed end-product (the links) with the raw data necessary to form the links. Twitter very definitely has some if not most of the information needed to create these links, and quite possibly has done some work in inferring connections, but are understandably very reluctant to share this work. Stating “the reason why we had not previously produced it because it’s not a category of information that we actually possess” seems perilously close to perjury for me, but then I’m a data manager and not a lawyer.
Extracting the raw data from Twitter’s servers is most likely a pretty easy task for qualified personell, but I understand that this transcript is from around february 2023, by which point they were severly short-staffed on the techincal side. Holtzblatt alludes to this when they “had an engineer working through the night” it most likely means he worked overtime.
The short version is: Twitter has streamlined complying with subpoenas for the most commonly sought information in a law enforcement portal. The data requested in this instance goes outside what is normally delivered but should still be easy to extract, but it looks like Twitter no longer has the competence or capacity to perform this work.