Book Review: Playing to the Edge

Click here to skip explanation/justification and to just get to the point…

I recently finished reading the book “Playing to the Edge: American Intelligence in the Age of Terror” by Michael Hayden, the former Director of the CIA and NSA.

I had a few goals that pushed me into reading this book. I had heard it mentioned on a podcast where Sam Harris interviewed Michael Hayden, and it sounded like a book where I could probably improve my world view of what goes on inside government intelligence agencies (though through a biased lens which would likely paint the NSA and CIA in a very favorable way). I wanted a sense of how agencies handling strategic issues make decisions and where there are levers of potential positive influence. It seems useful to know how incentives can be improved so that nothing bad happens if bad people are put into power or so that nothing bad happens if greatly expanding the power of the intelligence community becomes necessary for the preemption of catastrophic risks.

But I also had biases pushing me into reading the book: for a long while I have seen repeatedly false and often self-promotional claims by social commentators like Glenn Greenwald and I desired more ammunition to slow down anti-intelligence group-think I seem to experience online. Almost every time I read something from the Intercept, it doesn’t seem to take more than a few sentences to start finding claims which, with relevant background, are obviously false, but without that background are good for building a narrative with the readers about the police/surveillance state.  Though a lot of Greenwald’s analysis is solid, the pattern of pointing to all sorts of bad things the U.S. is doing or has done without proposing alternatives, or while ignoring others who do the same but far worse, was getting old.

There was plenty of the book I didn’t like. Similar to how at the Air Force Academy I was very annoyed with Condoleezza Rice wasting her speech talking about football instead of sharing her depth of knowledge and expertise, Hayden seems to use a similar social tactic in this book talking about the Pittsburgh Steelers and religion.  I suspect this is just him trying very hard to relate to average Americans, conveying what his daily life was like, which is perfectly understandable, but it does take up a lot of space. Despite this, the book overall had a lot of insight that I think people should not ignore when trying to understand how governments act.

1: Spy agencies in democracies with free medias are under immense pressure to not lie (or get caught lying). 

  • If your government does not have control of the media, the media can be as friendly or adversarial as it wants, based on profit. If the FSB tells an obvious lie, RT isn’t going to say anything, but if someone from the CIA or NSA lies publicly the agencies may suffer greatly from unpopularity (losing the ability to hire great people) and then policy change (congressmen can potentially enhance their ability to get more votes by pushing for regulation of what the public views as an out of control police state).
  • Because the public is (justifiably) worried about the government being too powerful, individuals can greatly increase their status by being external whistleblowers (instead of internal), even when they do so dishonestly. Much of what Snowden leaked was pretty irrelevant to demonstrating that the NSA is spying on U.S. citizens or doing anything bad, and many of his leaks sync up with his acquiring protection from different countries: by pointing out a controversial program which confirmed public fears, Snowden got away with basically being a spy for Russia and China and is popular instead of being hated.
  • The media isn’t censored, there are thousands of employees at the CIA and NSA: secrets get out eventually, and this leads to longer-term honesty incentives for intelligence agencies in countries with free speech. (While in countries without that incentive structure you get governments supporting contradictory argument spam, aka: the “Firehose of Falsehood“).
  • U.S. intelligence agencies polygraph their employees frequently to try to create an internal trust mechanism without having to disclose anything classified (though there is debate about how scientific polygraphs actually are, I expect the way intelligence agencies use them increases their effectiveness via setting baseline reactions, checking for consistency in stories, and using information advantages against those being asked questions).
  • Whenever possible without revealing classified information, U.S. intelligence agencies will opt to neither confirm nor deny claims instead of responding positively or negatively.

2. There are probably a lot of reasonable politicians who act partisan purely to further their instrumental goals.

  • If you are a congressman on an intelligence committee, in many cases may be well informed, but if anything goes wrong there is often the incentive to act that such is not the case (falsely acting as though one was deceived by the CIA or NSA).
  • If you are a congressman and support everything you know about that is going on in intelligence, it may still make sense to publicly raise hell, because no one knows what you say behind closed doors, and playing into public narratives can help you get votes.

3. Not deciding how to interpret laws in the face of new technology leads to overreach, a loss of strategic advantage, or both. 

  • Cyber warfare is different from conventional warfare, and different regulations are required to mitigate collateral damage. It is easier to attack and break a system than to observe a system while in cyberspace.
  • Cyber spying is different from intruding into someone’s home: in giant data sets you can anonymize personal information and just screen for relevant activity… but this is a choice, and if there are not solid institutional incentives in line, you can have abuse.
  • If good policy is not set ahead of time, and something bad happens, public reaction will push for the state to be more powerful than is optimal: both security and liberty are lost.
  • If optimal policy is set, intelligence agencies should be riding at the very edge of what they are allowed to do. Not doing so increases the odds of missing something important, and increases the odds of policy being pushed away from the optimum later due to mistakes and the resulting political pressure.

4. Public vindication of wrongdoing, proof of honesty, and proof of competence by intelligence agencies can sacrifice strategic advantage if brought forth too fast. 

  • In the process of proving innocence it is likely you have to explain a lot about what an agent or analyst is doing, which can reveal a lot of classified information.
  • If the U.S. determines something via spying, explaining the entire process to demonstrate honesty may make it impossible to gain information in the same way in the future, especially with cyber espionage.
  • Due to intelligence sharing, operations that seem to be from one country may actually be from another which really does not want to disclose its abilities to non-cooperating countries.

5. Because the media doesn’t know what is going on, it also doesn’t know what information causes a loss of significant strategic advantage.

  • This can be seen with the media publishing of certain leaks by Snowden, and then later retreats by the same outlets that certain disclosures were not ethical.
  • This can also be seen in cases where finance tracking programs were revealed, only to have the same media organizations inquire about why the NSA could not track illicit cash transfers as easily outside the country in later years…
  • Since the free media mostly isn’t trying to assess intelligence agencies from a military perspective, it doesn’t always know what information will be relevant and helpful to enemies.
  • The press DOES, however, understand strategic advantage when its own interests are at stake (such as when journalists are held hostage by terrorists for example) and is much better at delaying leaks in order to prevent death in such cases.

6. There are forms of torture that probably “work” (though no one wants to define themselves as torturing people).

  • While people may tell lies when put in pain, statements from different prisoners can be compared, cognitive availability can be taken account of, and terrorists have no clue how much intelligence interrogators do or do not have  (interrogators can directly test honesty for many statements which the captive does not have reason to believe the interrogator will know). Because of the information advantage of interrogators, getting people who are entirely non-compliant to talk at all often provides useful information.
  • There are religious extremists who claimed they were happy they were tortured: it was their belief that they would be punished (by God) if they shared any information without being tortured. Such people had a belief that it was morally permissible to share information following unbearable situations (if you believe the records on that), and they became very compliant with the CIA, verifiably answering many questions truthfully and adding valuable intelligence (once again, if you believe the records on that, people want to cover themselves and aren’t going to give all the details for even for self-vindicating claims for years due to controversy and need to protect strategic advantage).

7. The ethics of torture are a lot weirder than they publicly seem

  • (See the second point of 6 above about people being happy they were tortured…)
  •  Agency avoidance of public communication about enhanced interrogation makes sense not only because it looks bad, but also due to the risk of revealing limits that enable potential captives to train resistance exactly to the level required to withstand interrogation. (One can also imagine a world where countries/intelligence agencies act like they torture people very badly, but never do it since the fear of such pain causes people to talk before even mild pain is inflicted). Since waterboarding was revealed, if the U.S. ever goes back to using similar techniques on people they are likely to use worse ones due to trained resistance to what was made public.
  • Getting rid of enhanced interrogation reduced the willingness of the U.S. to actively take prisoners: we often opt to strike know terrorist instead of taking the risk to capture them and acquire information (I am pretty sure getting blown up is worse than getting waterboarded). This theme also gets brought up in a thought experiment by Sam Harris about the ethics of permitting any collateral damage vs. torture of combatants.
  • The media has probably waterboarded more people in protesting waterboarding than the CIA has actually waterboarded people.
  • The purpose of a lot of black-sites was basically just to deal with uncooperative countries making it hard to arrest terrorists (if it was about torture, they’d just fly everyone to the same country with the laxest laws + you really don’t want to release enemy combatants in the middle of a war, even if they are in a country where there is not currently fighting).

8. The enemy of my enemy is my friend… sometimes (people/media are bad at understanding multi-faction conflicts).

  • An intuition you can find from playing real-time strategy games, Risk,  reading the game theory example of a “truel,” or reading this book is that a lot of conflicts can last a long time if there are more than 2 factions. Since whoever strikes first either loses advantage to the factions that don’t have to fight or must engage multiple opponents at once, it makes sense to strive to be one of the factions initially uninvolved in conflict and to shift one’s position to force enemies to fight each other instead of you.
  • Example 1: Pakistan is not unambiguously supportive of terrorists, they just really hate India, and if they just let various militias control certain areas and attack the Indians, Pakistan gets the effect of India being attacked without provoking nuclear response and gets lot of people it doesn’t like to die doing such, rather than losing its own troops. Of course if you get too friendly with the minor enemy, eventually stuff gets a bit out of hand and Osama turns out to be next to your military academy…
  • Example 2: Syria has many factions fighting, and you can end up with weird situations like enemies trading fuel and weapons due to incentives to not fight each other. If ISIS gets totally wiped out first, Assad won’t seem as legitimate fighting the rest of the rebels (a lot of the “bad guys” will not seem as bad). If Assad can keep ISIS bordering rebels instead of his own troops, he can have the rebels being undermined from two directions and only have to focus on one opponent at time (in a given area).

9. Stopping Iran from getting nuclear weapons isn’t just about Israel and non-proliferation.

  • Iran already funds plenty of terrorism, if they have a successful nuclear weapons program, they could fund a lot more since they won’t be nearly as able to be invaded in retaliation.
  • For the foreseeable future, Israeli and U.S. missile defenses will probably stay ahead of Iran’s capabilities (how convenient, the U.S. has bases all around Iran! >_<), but this doesn’t stop weapons from being smuggled (though the consequences to using a nuclear weapon for Iran would probably be too great for Iran to risk that other than for religious extremist reasons.)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s