jwtadvice

The interagency here just means getting together the DoD, the White House, the BBG, USAID, CIA, etc to coordinate messaging so that they are all giving the same story.

Actually, the interagency is a catchall term to mean communication between different parts of the executive branch, as they otherwise report up to different Secretaries and command structures.

There are a number of different manners and ways that interagency coordination and collaboration is done and it is not specific to public messaging.

I'm curious how they handle the challenge they specified:

"Another challenge was building an AI that knows the difference between a fact and an opinion so that we do not flag opinions as false, since only facts can be false."

In news media that is typically considered credible, opinions are very often presented as fact, often marked with "an unnamed official said" or "our anonymous source told us."

Much of the news regarding the outside world is gathered at the State Department Press Briefings, where the Press Secretary will him/herself admit that what's being communicated at the stand is the opinion of the US government, after it goes through an interagency research, vetting, and message coordination phases.

Very often these coordinated opinions are then presented as fact at CNN, the New York Times, the Wall Street Journal and others.

Of course, this relationship led Americans to think that Saddam Housein had something to do with 9/11 and justified the invasion of Iraq, led the media to report on Mass Surveillance with the opinion label "Bulk Collection", led to inaccurate coverage of Bahrain, Kosovo, Benghazi, Jessica Lynch, etc. President Obama's National Security Advisor publicly bragged about the ability of coordination within the interagency to create an echo chamber in the US press.

I don't think there is an algorithm that can tell the difference between fact an opinion, unless somehow that algorithm is an automated investigative reporter.

It seems insurmountable at this stage of technology to try to separate US government propaganda or propaganda from interest groups from facts.

I hadn't considered it. The HN crowd is a lot better at understanding that downvotes are for comments that do not contribute to the conversation, rather than what they disagree with.

Maybe that's a more popular trend now? Having been active on HN for a long time and watched it evolve I guess I could see that...

Anyway, the branch of the conversation has gotten quite deep down the 'meta' rabbithole.

I wasn't certain, but it was the only grammatical error I could see in the comment.

Great resources!

It's important to understand that these approaches are on the leading edge of research, and represent unsolved problems!

A bit ridiculous that I'm being downvoted for misspelling "its" as "it's". I've now corrected it in the above post.

The media industry in the United States, primarily because it functions as a propaganda medium for state narrative, has increasingly been losing credibility with its audience.

The 'fake news' thing, the 'troll' thing before that, the 'political spam' thing before that: these are attempts to prevent the manifestations of credible alternatives that can attract funding, challenge the doctrinal ideology espoused by the network of media, or further disrupt the reputation of the current infrastructure for engaging in public diplomacy.

The problem is much more difficult than creating a classifier. The purpose of filtering news is to control the impact, not to control the tactic. The last time Facebook issued a surge to control 'spam' posts for the same government concerns they seek to deal with "fake news" today, they took out pro-Sanders articles, censored sharing of the Snowden documents, prevented people from linking to Wikileaks, and shut down Facebook organizing for May Day protests.

Facebook regularly censors and directs conversation online around the world to the whims of various regimes and administrations, but it's just not easy "use machine learning".

The issue is that machine learning is inherently modeled to solve decision problems in spaces that separate high level 'features' from low level observations. The study and practice of machine learning thus far has modeled noise, uncertainty and complex state spaces.

However, machine learning classifiers do not themselves have a concept of adversarial data: this led, for example, to spammers who would "poison" machine learning classifiers by biasing them to learn the wrong things, so that they could get their spam through. Similarly, not too long ago, Microsoft put a conversation bot on the web that 'learned the wrong things' because people approached the bot in an attempt to trick it.

Facebook's requirement for battling unsanctioned opinions is that it solves this problem - it hasn't been solved yet. Facebook needs to continuously monitor the adapting activity and patterns of people who are trying to say things that governments and Facebook customers don't want people saying.

Basically: machine learning doesn't adapt but people do, and naive attempts to get machine learning to adapt make it vulnerable to being tricked, because classifier theory inherently has no concept of adversarial data.

The argument at the top of this article DEEPLY confuses the issue of mass surveillance, which I think is a very bad way to start an article titled “Time for a Rigorous National Debate about Surveillance”.

The article begins by invoking the spectral boogyman of terrorism, and doubles down on the concept that the United States can somehow be at war with a tactic.

The article refuses to admit that mass surveillance, as disclosed by the Snowden Documents, revealed that the capabilities of ubiquitous espionage were not developed and are not primarily used for a ‘war on a tactic’.

What the disclosures and investigative journalism clarified is that mass surveillance is about power and control writ large, with surveillance capabilities having been build long before terrorism became relevant and used at that time for the same power and control purposes that it is used today.

Specifically, the disclosures revealed that the intelligence is used to understand populations and how information moves through them - this helps with propaganda operations in various settings. They also revealed that the espionage was used to aide US diplomats by giving them negotiating leverage in international meetings, and was used to support domestic industries by stealing relevant market information from overseas competitors.

This slight of hand - equivocating public safety with national security - is a common tactic. There is a very specific definition of national security, defined by National Security Presidential Directive-1 (http://fas.org/irp/offdocs/nspd/nspd-1.htm): "National security includes the defense of the United States of America, protection of our constitutional system of government, and the advancement of United States interests around the globe. National security also depends on America's opportunity to prosper in the world economy.”

As the homeland is the safest place on Earth, the typical invocation of "national security" refers to the other components: a prospering economy, overseas access and interests, and protection of the current global political system for which America sits on top.

Mass Surveillance helps America to maintain it’s edge as a leading world power.

It is not about boogymen and terrorism, or about defeating certain kinds of tactics.

The article is completely ingenuous.

The same kinds of leaks that were problematic for the last administration are problematic for the new administration. There's a small reset of credibility with the new administration - but not much.

The strategy is highly likely to be: magnify the idea that the last administration was corrupt, and suppress the idea that the new administration is corrupt (by suppressing whistleblowing, etc).

It's not a good argument because none of the Snowden Documents put individuals at risk. What he did do was disclose massive illegal and illegitimate activity, promulgating awareness of a giant surveillance network with built in parallel construction for the purposes of diplomatic and industrial espionage and mass propaganda.

That put our "military in danger" the same way setting innocent victims of military torture free: yeah, now there are people who are pissed off with legitimate gripes.

All of your data from the smart phone service, however, is still going to be available for government surveillance purposes. This makes me unlikely to buy this product - from a reseller or not.

> Nothing listed is a backdoor.

I'm baffled. I like to think of the quality of HN comments as much higher.

Apple had options to make this data unavailable by design to themselves and to law enforcement. They chose a design so that they could provide this information. The phone does it without the user's consent.

However, I understand that there are people who would rather redefine terminology to suit their cognitive dissonance.

In any case we can agree on the following:

-----------

Apple provides a near majority of your sensitive information to law enforcement by the design of the product, and you can not use the product in a meaningful way without that information becoming available to state surveillance and state law enforcement.

> Timestamps and durations have been a part of phone UIs for quite some time

This is not familiar to me at all. If true, note the remaining points: the manner in which this data was synchronized allowed this material to be provided to law enforcement (Apple had other options) and Apple also included information from third party communication tools used by some people (mistakenly) to avoid being surveilled.

Yes.

For an understanding of what data Apple gives surveillance operations access to, the following list is enumerated in an older Law Enforcement Access document: https://news.ycombinator.com/item?id=12983081

It's also important to understand that the purpose for this surveillance (this following section being informed by the Snowden Disclosures) is not merely for investigative work or for tracking down terrorists. Intelligence work of this kind is used in HUMINT operations (mass propaganda), for industrial espionage, for diplomatic espionage and for signals collections (intelligence used to break into networks).

It is not clear whether Apple shares these concerns with the US government, or whether it shares lesser concerns that show up at a much lower frequency.

For those that don't understand this comment, intelligence agencies are known for generating fake evidence and allegations to take down targets - and it's well known that intelligence agencies such as the NSA and GCHQ are not charged with missions to track down terrorists.

A very clever sarcastic comment I think will probably go under the radar because the OP omitted a "/s".

I agree with this. Government leverage on Apple is huge, as is the magnitude of public-private partnership. Mass surveillance of its products is not likely something that Apple would have pursued on its own.

However, it Apple has capitulated to a combination of carrots and sticks. Today, it enables surveillance of its customers.

> Nice job slipping a completely unfounded lie into your response.

Oh my.

Yes Apple backdoors their phones for government surveillance access.

There's no sneaking here. I'm saying it very clearly as a central tenet of my comment.

-----------------------

Here's a list of things that Apple will provide from Apple's own Guide for Law Enforcement Access:

- Device Registration (name, address, email address, telephone number, iCloud Apple ID)

- Customer Service Records

- iTunes (name, physical address, email address, and telephone number, purchase/download transactions and connections, update/re-download connections, and iTunes Match connections, iTunes subscriber information and connection logs with IP addresses, specific content purchased or downloaded).

- Apple Retail Store Transactions (cash, credit/debit card, or gift card transactions, type of card, name of the purchaser, email address, date/time of the transaction, amount of the transaction, and store location, receipt number)

- Apple Online Store Purchases (name, shipping address, telephone number, email address, product purchased, purchase amount)

- iTunes Gift Cards (sixteen-digit alphanumeric code, nineteen-digit code, any purchases, name of the store, location, date, and time, user account

- iCloud (music, photos, documents, iCloud email, encryption keys, Subscriber Information, iCloud feature connections, connection logs with IP addresses, Mail Logs, records of incoming and outgoing communications such as time, date, sender email addresses, and recipient email addresses, Email Content, Other iCloud Content, Photo Stream, Docs, Contacts, Calendars, Bookmarks, iOS Device Backups, stored photos, documents, contacts, calendars, bookmarks and iOS device backups, photos and videos in the users’ camera roll, device settings, app data, iMessage, SMS, and MMS messages and voicemail)

- Find My iPhone (including connection logs)

- Other Available Device Information (MAC Address for Bluetooth, Ethernet, WiFi, or FireWire)

- Requests for Apple Retail Store Surveillance Videos

- Game Center (Connection logs with IP addresses, specific game(s) played)

- iOS Device Activation (including upgrades the software, IP addresses, ICCID numbers, and other device identifiers)

- Sign-on Logs (iTunes, iCloud, My Apple ID, and Apple Discussions, Connection logs with IP addresses, Sign-on transactional records)

- My Apple ID and iForgot Logs (password reset actions, Connection logs with IP addresses)

- FaceTime (logs when a FaceTime call invitation is initiated, content protected by 15 bits of entropy if secure enclave baked key is obtained from manufacturer)

And, from this thread on HN today (https://news.ycombinator.com/item?id=12977612): All call logs including contacts, timestamps, and durations including for third party applications on the phone like WhatsApp, Skype and Viber.

The "secret" part is that they include information that is only interesting to law enforcement (timestamps and durations) in addition to pulling information from third part communication tools like WhatsApp.

Namely, Apple chose to provide this information to government surveillance when they could have (and indeed promised) to design it in a way that allows for the iPhone to be used in sensitive situations and by people who need security from state coercion and violence.

Eugene.

We agree about the need to use encryption.

We disagree about it's limitations, and how to use it effectively.

Unfortunately this isn't a theoretical debate. Journalists, for example, around the world are being killed at increasing rates by governments, with this year reaching an all time high. State surveillance is an issue of tremendous importance for civil rights and personal freedoms. Our conversation and recommendations have to be sober and give candid, informed advice.

I think you absolutely have the right motivations: to encourage people to protect themselves from state coercion and violence.

Sure thing!

Unfortunately I'm afraid that you missed a lot of the content! You had asked for specific documents, not news reporting that pieced the context together (for example following the code names of the various programs together to understand how the system works as a pipeline).

In doing so, I'm afraid you're deeply underestimating the capabilities of the NSA, which expert analysis of the documents in question along with the technologies, related programs and internal customer requests indicate something much more startling than Debian session issues.

To put a point on that: TLS is broken at scale for billions of internet browsing sessions across platforms.

Give the short amount of time you've looked at the documents, I understand the misinterpretation you've arrived at. At the same time it's really impressive that you bothered to try to read the documents at all and should be commended.

Please feel encouraged to continue reading.

A large number of documents are hosted on edwardsnowden dot com and search functionality is provided here: https://search.edwardsnowden.com/

Der Spiegel, The Guardian and The Intercept provided good analysis of the documents as they were being released and that can be found at the websites respectively.

> It's sad, but I'm still saying "use encryption", because alternative is "don't use encryption" and it's obviously worse. So I'm much more responsible than those who are trying to propose second option.

Yes. I agree with this. Just don't recommend this as 'enough' to people who have serious need to protect their communications. If someone is worried about their porn habits, sure. But recommending this for sensitive use (journalists, dissidents, organizers, politicians, administrators) can lead to very serious outcomes.

For these cases, it is important to have frank conversations about the limitations of freely available and widely distributed tools.

You are following my logic correctly.

To be fair, the amount of leverage the US exerts on businesses for military purposes is astounding. This should be factored into the understanding.

I understand that you wrote what you wanted to write.

Unfortunately what you wrote was FUD in the worst case and misinformed in the best.

Anyway I think we both are at a point where we understand the other's point of view. Thank you for the conversation. I believe we both hope a lot of HN folks will stumble over it over the next few days. :)

I used the term 'motivate' specifically because of the PR language intending for customers to evaluate the Apple product as something that could be used by those who need to use their phone for private and/or sensitive reasons.

Apple of course backdoors their phones for government surveillance access. But they do motivate a threat model that includes government surveillance.

I know parsing my comment in this way may seem difficult, but I used the terminology I did on purpose.

There are no illusions that Apple achieved the security properties that it has motivated.

Google Pixel does not even pretend to address the security concerns of journalists, politically active citizens, IT professionals, or individuals contacting attorneys.

The two comments have significantly different content.

For those curious here is the other comment in question: https://news.ycombinator.com/item?id=12982502

The thing to note here is that it includes all information about the calls including the times and durations and the logs also intercept communications from third party apps like Skype, WhatsApp, and Viber, and that all of this information is available to law enforcement.

This is despite Apple's PR statements claiming that the company has designed the phone and its logs to minimize or eliminate the amount of information will be used to feed into law enforcement surveillance requests.

Given recent news about mass surveillance, it's important to note that Pixel's security model does not and can not seek to protect your data for use for private messaging, conversation with attorneys, for journalists, or to organize for political reasons.

If you are interested in a communications device that can be used for any of these things, Pixel's security model will not cover you and you will need to look for an alternative product.

If you go down that route, convincing the public to use and standardizing backdoored and broken cryptography would also count as social engineering.

The communications are decrypted.

This is done through a combination of backdoored cryptography, state of the art cryptanalysis, and mass social engineering (propaganda).

Namely, you're not proposing something that's mutually exclusive to decryption of data.

Sure, there's been good journalist work compiling the state of the art here: http://www.spiegel.de/international/world/nsa-documents-atta...

But also, being familiar with the cryptography you are attempting to have conversation about is a basic starter. For example, there were attacks listed in the prior comment following the "oh there's a ton" header. You seemed to have glazed past those, but they are familiar topics to those who habituate themselves with the technicalities of cryptography and its practice.

Pixel/Google does not motivate a threat model under which to evaluate or understand their design and marketing promises, but we can take a hint from "protects your data if your phone falls into someone else's hands." - Namely thefts of opportunity.

Unlike other phone manufacturers, Google does not promise potential customers that your data will be protected from Google, it's partners and from law enforcement and mass surveillance programmes.

Therein this product doesn't provide a stronger security posture that competitors - and furthermore it's threat model and security properties do not meet what are in my opinion minimal reasonable requirements.