One of the reasons to get DirecTV out of AT&T was to run it like a business, which AT&T couldn't do.
TPG ditched OANN, like private equity companies do, closing financially "underperforming" news media. I didn't put "news" in quotes there because a PE's spreadsheets can't tell the difference between Chanel Rion and real reporters. DirecTV will likely lose the My Pillow infomercial channel as a result of this decision -- not much revenue lost there with the My Pillow brand imploding.
... and disses everyone else with the same member name (something to do with Anonymous Coward)? Is there a locus classicus for the participants in this naming dispute so the rest of us could engage in discussion? Mods, could you issue these guys sequence numbers or something?
the DOJ, at least when I was there, acted as if the trust of the courts was its most valued asset. It's not an asset you buy cheap.
The AG's memo of last July says "[t]he Department of Justice will no longer use compulsory legal process for the purpose of obtaining information from or records of members of the news media acting within the scope of newsgathering activities."
If I follow this, they are saying that they will not ask the court for a warrant without making the determination that there was an activity outside the scope of newsgathering activities. And this court wants to hear about that determination. (Lots of formal statements of prosecutorial discretion policy drift toward regulating DOJ/court interactions as if they were law.)
So, e.g., if they are asking for an arrest warrant for an act which is illegal and falls outside the scope of newsgathering, this is conceptually easy -- so if they are accusing Mr. Shroyer of a violent act, or of interfering with an official proceeding, might be easy. If they are accusing him of something journalists do all the time -- failing to disperse or taking photos of the police pushing the crowd -- also conceptually easy, within the scope of newsgathering. Who has any doubt that a difficult border of "scope of newsgathering activities" will arise in some future matter?
Shroyer appears to have declared himself to be part of the event not part of the coverage of the event, so it seems the DOJ could easily meet its own test. But that's not all that's needed, they also need to leave courts with a high level of trust that they did meet their own test.
The multiband nature of 5G standards has been out there for over a decade. I first heard of it in 2008, and I don't do this every day. Interested parties, such as the airline industry, weighed in on the FCC rules for it for years. The FAA now thinks further testing and adjustment is needed. Where were they when the issues were being decided? To come in now, after many billions of dollars have been spent in reliance on the regs, is not good government. There is an orderly process for interagency communication about future decisions to avoid messes like this. DOD and DOT are the least cooperative, in my experience, and through delay, the least responsible -- soldiers will die if I don't get to decide your policy for you, and planes will fall from the sky. No details needed, just the assertion that planes might fall, coming in decades late.
This is a broad general problem for automatic/algorithmic flagging, which the FB guys are farther down the road of understanding than the twitter folks because they got burned on it earlier. An automatic flagging system predicts whether there is a policy violation based on observables (including, e.g., photo content, automatic systems can be super powerful.) As long as that relationship is stable, the system will work. When a hostile actor can change that relationship, either by trying to change it or just by changing behavior, it blows up.
Why don't engineers quickly tumble to this? It is a human behavior problem, not a technical one.
I bought an ifit treadmill. The physical machine is moderately cool, better than the ones at my gym (where I didn't go for some years, pandemic.) The software and systems and the organization are awful. Long workouts are tough, since the s/w crashes regularly. I now have an intervals workout on the machine -- I did manual intervals and go back to the calendar to re-use that old workout. It has built in music, but after a couple of weeks I put an Alexa next to it, much much better. Tech support -- completely incompetent. Service and support -- two year delays to get a bolt back in, but there is a thriving if very expensive third party entrepreneurial repair sector. There must be more ...
... Any of you techdirt readers have money as well as the brilliance we all love here, buy ifit, bring it up to the organizational competence level of, say, Sprint just before it was bought, and you will make a ton of $$.
Reporters can struggle to understand the documents they are reading and excerpting, especially technical documents. Publishing the originals will let alternative, sometimes better, understandings emerge. Thank you to Gizmodo.
Like most lawsuits from Trumpland, the point of this one is not ... how to explain the scope of the "not" ... not anything to do with winning a lawsuit or what the law might let one enforce. Not anything to do with the WaPo or the NYT's reporting or with their Pulitzer Prize. Not anything to do with the events it says it is about.
The letter, and the lawsuit if one follows, are PR to be lapped up by people who are already convinced that "the mainstream media" has zero standards, lies all the time, and that the only source of truth is DJT who is Fighting For You. The letter has no other function. Ahh, that "no" is the scope of the "not."
The main confusions in this bill appear to be
Attempting to regulate what a class of algorithms does by limiting its access to automatically collected data.
No clean definition of ranking, the regulated object.
Automatic collection of information equated to lack of transparency
also equated to lack of user buy-in or consent
There must be more, but that's enough to demonstrate the main point of the post, that this is an extremely confused law. That's even before you get to the sensible points that automatic data collection can be a wonderful convenience and that the automatic collection bears no logical connection to the goals of the algorithm, good bad or indifferent, other than making them easier to attain.
Is there an example, better yet many examples, of speedy and effective enforcement of any copyleft license against commercial entities?
Artifex seems to have had some good luck in the courts after many years of work. I mention Artifex mostly to point out that a search by an amateur like me doesn't find much, perhaps some of the real experts on techdirt know the case law :)
Did the police search the car by intrusion? The dog's "instincts" don't seem to be a valid reason to say there was not police search nor a valid reason to say there was no intrusion.
James Thurber updated: If you are a police dog, where's your nose?
... of the nonsense politicians pull when they yell "fake news." Thugs assaulting the US capitol? Patriots! Tourists! Other thugs roughing up school boards? Concerned parents! Idiots who deny the existence of communicable disease? Guardians of religious freedom! Those were comparatively heavy lifts of nonsense.
At least understanding that, having stumbled upon a security leak immediately telling the leaky site and then only later publishing news about is a good thing and is pro-security is very difficult. Wait, sorry, it is not very difficult.
Automatic filtering to find prohibited copying cannot work. 1) As many have pointed out, much copying is protected under the copyright laws (parody is an example). Automatic filtering systems may be improved to where they can detect a copied melody, but they will not detect parody even if there are dramatic improvements in AI. 2) Many copies are secretly posted by marketers working for the copyright holder to build social media buzz. Automatic filters can't detect that either.
Why do copyright holders want a system that will fail? These guys are the champions of nuisance lawsuits. They like hard to prove causes of action.
in this instance, there is no political consensus to pass any specific reform, since the near consensus that Section 230 creates "a problem" hides fundamental disagreements about what the problem might be, with such common gripes as "too much disinformation" and "they are censoring discriminatorily" calling for changes in opposite directions.
Thank you for this insightful article, Ms. Gellis.
Thanks, R.H. Your second paragraph seems accurate, but to me it appears just to repeat what is funny about the doctrine. A private search has not occurred if a private person just told a computer -- not a person -- to look for the evidence and tell the police if any is found.
Perhaps the key is that the "just have to ask for a warrant" step prevents the oddity of the definition from doing any very large harm in practice.
The ICAC's failure to apply for a warrant it would easily get seems like a basic error. So the "cops" part of this is easy. But the "courts" part is not so easy.
The courts are drawing a funny distinction. "No one at Google had opened or viewed Wilson’s email attachments; its report was based on an automated assessment that the images Wilson uploaded were the same as images other Google employees had earlier viewed and classified as child pornography." The appeals court makes the same distinction " the government agent viewed Wilson’s email attachments even though no Google employee—or other person—had done so, thereby exceeding any earlier privacy intrusion."
So what are they going to do when some of these scum start changing the image file just a little bit to escape the hash check, and the image is identified by a machine learning algorithm as extremely likely -- for you who aren't probabalists, significantly more likely than that this "e" arose because I typed the letter e -- to be child pornography. (I won't say "AI" because classifying the machine learning algorithm as a person offends common sense.) The right answer seems to be "grant the warrant" and perhaps that is not inconsistent with what these courts found.
But the distinction is funny in another way. There was "no earlier privacy intrusion" when a computer at Google looked at the file and reported it? Whoa! Google "knows" a lot about me, meaning that their computers do. No privacy intrusion until a human looks at it? Maybe "privacy intrusion" is a very limited term of art, but the principle seems funny.
Where I am coming from here is that almost nothing Google does is done by humans; not suggesting a response to a search, not suggesting an ad for the ad auction associated with the search, not flagging an image as child pornography. Almost everything is done by machines, and what the machines do might be reviewed humans very infrequently. So it seems like a funny distinction.
Reporters, like civilians generally, struggle to make sense of what courts do. My daughter, who took tons of journalism classes, became extremely educated in some aspects of the First Amendment, e.g. about lawsuits against journalists, but didn't much learn about how courts work.
Could there be a simple tutorial for journalism students? Many reporters took those courses and wrote for the college paper. Journalists' biggest problem is assuming that every court ruling must be substantive, when most are procedural. The confusion in the Trump seems like one that a tutorial for journalists could really help with and that would also help with the biggest problem. It could teach them what are the basic steps of a matter, filing, pretrial motions, injunctions vs final resolution, discovery, dismissal vs settlement trial, etc. enough so that they would be equipped to start by asking what the event of news they are starting to write about might be.
I'd also say they could read these wonderful discussions to learn more, but I suspect most journalism students come with fixed opinions about what the courts should do, being young.
just an ordinary business decision
One of the reasons to get DirecTV out of AT&T was to run it like a business, which AT&T couldn't do.
TPG ditched OANN, like private equity companies do, closing financially "underperforming" news media. I didn't put "news" in quotes there because a PE's spreadsheets can't tell the difference between Chanel Rion and real reporters. DirecTV will likely lose the My Pillow infomercial channel as a result of this decision -- not much revenue lost there with the My Pillow brand imploding.
thank you
Re: Re: so everybody has the same member name ....
Okay, thanks. You are Purple AC.
so everybody has the same member name ....
... and disses everyone else with the same member name (something to do with Anonymous Coward)? Is there a locus classicus for the participants in this naming dispute so the rest of us could engage in discussion? Mods, could you issue these guys sequence numbers or something?
prosecutorial discretion and the trust of the courts
the DOJ, at least when I was there, acted as if the trust of the courts was its most valued asset. It's not an asset you buy cheap.
The AG's memo of last July says "[t]he Department of Justice will no longer use compulsory legal process for the purpose of obtaining information from or records of members of the news media acting within the scope of newsgathering activities."
If I follow this, they are saying that they will not ask the court for a warrant without making the determination that there was an activity outside the scope of newsgathering activities. And this court wants to hear about that determination. (Lots of formal statements of prosecutorial discretion policy drift toward regulating DOJ/court interactions as if they were law.)
So, e.g., if they are asking for an arrest warrant for an act which is illegal and falls outside the scope of newsgathering, this is conceptually easy -- so if they are accusing Mr. Shroyer of a violent act, or of interfering with an official proceeding, might be easy. If they are accusing him of something journalists do all the time -- failing to disperse or taking photos of the police pushing the crowd -- also conceptually easy, within the scope of newsgathering. Who has any doubt that a difficult border of "scope of newsgathering activities" will arise in some future matter?
Shroyer appears to have declared himself to be part of the event not part of the coverage of the event, so it seems the DOJ could easily meet its own test. But that's not all that's needed, they also need to leave courts with a high level of trust that they did meet their own test.
time is needed for testing and making planes safe: 2008-2019
The multiband nature of 5G standards has been out there for over a decade. I first heard of it in 2008, and I don't do this every day. Interested parties, such as the airline industry, weighed in on the FCC rules for it for years. The FAA now thinks further testing and adjustment is needed. Where were they when the issues were being decided? To come in now, after many billions of dollars have been spent in reliance on the regs, is not good government. There is an orderly process for interagency communication about future decisions to avoid messes like this. DOD and DOT are the least cooperative, in my experience, and through delay, the least responsible -- soldiers will die if I don't get to decide your policy for you, and planes will fall from the sky. No details needed, just the assertion that planes might fall, coming in decades late.
hard to do automatic flagging when behavior changes
This is a broad general problem for automatic/algorithmic flagging, which the FB guys are farther down the road of understanding than the twitter folks because they got burned on it earlier. An automatic flagging system predicts whether there is a policy violation based on observables (including, e.g., photo content, automatic systems can be super powerful.) As long as that relationship is stable, the system will work. When a hostile actor can change that relationship, either by trying to change it or just by changing behavior, it blows up.
Why don't engineers quickly tumble to this? It is a human behavior problem, not a technical one.
incompetence is often a better explanation than ill will
I bought an ifit treadmill. The physical machine is moderately cool, better than the ones at my gym (where I didn't go for some years, pandemic.) The software and systems and the organization are awful. Long workouts are tough, since the s/w crashes regularly. I now have an intervals workout on the machine -- I did manual intervals and go back to the calendar to re-use that old workout. It has built in music, but after a couple of weeks I put an Alexa next to it, much much better. Tech support -- completely incompetent. Service and support -- two year delays to get a bolt back in, but there is a thriving if very expensive third party entrepreneurial repair sector. There must be more ...
... Any of you techdirt readers have money as well as the brilliance we all love here, buy ifit, bring it up to the organizational competence level of, say, Sprint just before it was bought, and you will make a ton of $$.
openness is better here
Reporters can struggle to understand the documents they are reading and excerpting, especially technical documents. Publishing the originals will let alternative, sometimes better, understandings emerge. Thank you to Gizmodo.
from the dude,-come-on dept
Like most lawsuits from Trumpland, the point of this one is not ... how to explain the scope of the "not" ... not anything to do with winning a lawsuit or what the law might let one enforce. Not anything to do with the WaPo or the NYT's reporting or with their Pulitzer Prize. Not anything to do with the events it says it is about.
The letter, and the lawsuit if one follows, are PR to be lapped up by people who are already convinced that "the mainstream media" has zero standards, lies all the time, and that the only source of truth is DJT who is Fighting For You. The letter has no other function. Ahh, that "no" is the scope of the "not."
The main confusions
The main confusions in this bill appear to be
Attempting to regulate what a class of algorithms does by limiting its access to automatically collected data.
No clean definition of ranking, the regulated object.
Automatic collection of information equated to lack of transparency
also equated to lack of user buy-in or consent
There must be more, but that's enough to demonstrate the main point of the post, that this is an extremely confused law. That's even before you get to the sensible points that automatic data collection can be a wonderful convenience and that the automatic collection bears no logical connection to the goals of the algorithm, good bad or indifferent, other than making them easier to attain.
trump given 30 days ...
Is there an example, better yet many examples, of speedy and effective enforcement of any copyleft license against commercial entities?
Artifex seems to have had some good luck in the courts after many years of work. I mention Artifex mostly to point out that a search by an amateur like me doesn't find much, perhaps some of the real experts on techdirt know the case law :)
is there any element of intent in a 4A violation?
Did the police search the car by intrusion? The dog's "instincts" don't seem to be a valid reason to say there was not police search nor a valid reason to say there was no intrusion.
James Thurber updated: If you are a police dog, where's your nose?
I am sad to report that this doesn't push the boundary ...
... of the nonsense politicians pull when they yell "fake news." Thugs assaulting the US capitol? Patriots! Tourists! Other thugs roughing up school boards? Concerned parents! Idiots who deny the existence of communicable disease? Guardians of religious freedom! Those were comparatively heavy lifts of nonsense.
At least understanding that, having stumbled upon a security leak immediately telling the leaky site and then only later publishing news about is a good thing and is pro-security is very difficult. Wait, sorry, it is not very difficult.
not possible this will work
Automatic filtering to find prohibited copying cannot work. 1) As many have pointed out, much copying is protected under the copyright laws (parody is an example). Automatic filtering systems may be improved to where they can detect a copied melody, but they will not detect parody even if there are dramatic improvements in AI. 2) Many copies are secretly posted by marketers working for the copyright holder to build social media buzz. Automatic filters can't detect that either.
Why do copyright holders want a system that will fail? These guys are the champions of nuisance lawsuits. They like hard to prove causes of action.
fortunately
in this instance, there is no political consensus to pass any specific reform, since the near consensus that Section 230 creates "a problem" hides fundamental disagreements about what the problem might be, with such common gripes as "too much disinformation" and "they are censoring discriminatorily" calling for changes in opposite directions.
Thank you for this insightful article, Ms. Gellis.
Re: Re: the courts are drawing a funny distinction
Thanks, R.H. Your second paragraph seems accurate, but to me it appears just to repeat what is funny about the doctrine. A private search has not occurred if a private person just told a computer -- not a person -- to look for the evidence and tell the police if any is found.
Perhaps the key is that the "just have to ask for a warrant" step prevents the oddity of the definition from doing any very large harm in practice.
the courts are drawing a funny distinction
The ICAC's failure to apply for a warrant it would easily get seems like a basic error. So the "cops" part of this is easy. But the "courts" part is not so easy.
The courts are drawing a funny distinction. "No one at Google had opened or viewed Wilson’s email attachments; its report was based on an automated assessment that the images Wilson uploaded were the same as images other Google employees had earlier viewed and classified as child pornography." The appeals court makes the same distinction " the government agent viewed Wilson’s email attachments even though no Google employee—or other person—had done so, thereby exceeding any earlier privacy intrusion."
So what are they going to do when some of these scum start changing the image file just a little bit to escape the hash check, and the image is identified by a machine learning algorithm as extremely likely -- for you who aren't probabalists, significantly more likely than that this "e" arose because I typed the letter e -- to be child pornography. (I won't say "AI" because classifying the machine learning algorithm as a person offends common sense.) The right answer seems to be "grant the warrant" and perhaps that is not inconsistent with what these courts found.
But the distinction is funny in another way. There was "no earlier privacy intrusion" when a computer at Google looked at the file and reported it? Whoa! Google "knows" a lot about me, meaning that their computers do. No privacy intrusion until a human looks at it? Maybe "privacy intrusion" is a very limited term of art, but the principle seems funny.
Where I am coming from here is that almost nothing Google does is done by humans; not suggesting a response to a search, not suggesting an ad for the ad auction associated with the search, not flagging an image as child pornography. Almost everything is done by machines, and what the machines do might be reviewed humans very infrequently. So it seems like a funny distinction.
"what the hell is wrong with reporters these days?"
Reporters, like civilians generally, struggle to make sense of what courts do. My daughter, who took tons of journalism classes, became extremely educated in some aspects of the First Amendment, e.g. about lawsuits against journalists, but didn't much learn about how courts work.
Could there be a simple tutorial for journalism students? Many reporters took those courses and wrote for the college paper. Journalists' biggest problem is assuming that every court ruling must be substantive, when most are procedural. The confusion in the Trump seems like one that a tutorial for journalists could really help with and that would also help with the biggest problem. It could teach them what are the basic steps of a matter, filing, pretrial motions, injunctions vs final resolution, discovery, dismissal vs settlement trial, etc. enough so that they would be equipped to start by asking what the event of news they are starting to write about might be.
I'd also say they could read these wonderful discussions to learn more, but I suspect most journalism students come with fixed opinions about what the courts should do, being young.
Re: Re: Re: Re: Re: geofence warrants
Thank you also mhajicek and DannyB, I learned a lot from this discussion as well.
More comments from freelunch >>
Techdirt has not posted any stories submitted by freelunch.
Submit a story now.
Tools & Services
TwitterFacebook
RSS
Podcast
Research & Reports
Company
About UsAdvertising Policies
Privacy
Contact
Help & FeedbackMedia Kit
Sponsor/Advertise
Submit a Story
More
Copia InstituteInsider Shop
Support Techdirt