February 16, 2010 at 10:51AM via Pocket
Internal government documents obtained by the Electronic Privacy Information Center have revealed that the US Department of Justice is secretly helping AT&T and other service providers evade wiretapping laws so that the US government can conduct surveillance on parts of their networks. The legal immunity comes from authorizations granted by the Justice Department through special “2511” letters that absolve carriers in the event that the surveillance is found to run afoul of federal law.
The authorization program began as a narrow cybersecurity effort to monitor government defense contractors, but has been expanded to cover critical infrastructure like energy, finance, and health care, CNET reports. Normally, the Wiretap Act prohibits…
Effective messaging is essential to a successful job search.
from Jim's Feeds - National News http://bit.ly/11Dh6nK
Cryptographers have once again put SSL/TLS (that’s the padlock in HTTPS) in their gunsights and opened fire.
This time, they’ve done some severe damage.
Paul Ducklin takes a detailed look…
from Naked Security - Sophos http://bit.ly/WvvXiJ
Ten years ago this week, the U.S. invaded Iraq, citing intelligence that turned out to be bogus. I had to work on some of it — and I also had to work on keeping the really, really terrible versions of it out of our analysis.
Specifically, I was a CIA analyst working in the Counterterrorism Center in the overburdened days after 9/11. As analysts, we spend most of our time identifying burgeoning issues based on communications intercepts, reports from CIA case officers, imagery from satellites, accounts from other governments, and piecing together a story.
What we don’t do routinely is tie one catastrophe to another. But that was exactly what I was asked to do in November 2002, shortly after Congress voted to authorize war with Iraq. That war was predicated on Saddam Hussein’s (ultimately nonexistent) stockpiles of deadly weapons, but lurking in the background was the assertion that he’d pass them on to al-Qaida. At the CIA’s Iraq Branch in the Counterterrorism Center, we didn’t think Saddam had any substantial ties to al-Qaida. But soon we found ourselves fielding questions from determined Bush administration officials about whether Saddam was tied to 9/11.
That’s how my team ended up in a windowless room with my branch chief, “Karen,” who was pretending to be Dick Cheney or his chief of staff, Scooter Libby.
That month, Vice President Cheney scheduled a meeting with our Branch to discuss our assessment of Iraq’s relationship with al-Qaida and 9/11. It was his second visit to the Branch; there always seemed to be more questions. The Branch Chief called us together for a practice session in a bland conference room a few days before their arrival. At this so-called “murderboard” session, we weren’t stripping down our analysis to find data we’d missed. We were practicing how to defend our perspective when questioned by the Vice President of the United States.
The Branch Chief would get the ball rolling with questions designed to lead us down a rabbit hole. Karen had briefed Libby, so she was skilled at impersonating both the Vice President and Libby — that is, she was being relentless and insistent — to anticipate the questions they would ask. We had a bottom line: Fear of Islamic extremism growing in Iraq would limit Saddam’s willingness to work with bin Laden. Fake-Cheney would rejoinder: Would ideological differences really hinder their cooperation? Anticipating the response, she’d come back with: What if bin Laden convinced Saddam that acting against the United States was in both of their best interests; you have told us we don’t know exactly how much communication has taken place between the regime and al-Qaida; and you have already found information that specified safe havens, contact and training?
We needed to poke holes in our analysis, to be sure we were right. If not, we could rest assured Cheney would. Already, Cheney’s Pentagon ally, Undersecretary of Defense Doug Feith, had put together an alternative analysis faulting our own and asserting instead that “multiple areas of cooperation” existed between al-Qaida and Saddam. The ongoing questions and briefings became a labyrinth.
How far down a rabbit hole should we go in answering questions? Will it be misconstrued as an actual answer based on a made-up scenario? It was an unorthodox practice. But we were unused to a senior political figure being willing to dig down into the details of our analysis.
In the abstract, challenging CIA’s analysis is a good thing: agency analysts get stuff wrong, as evidenced by Saddam’s non-WMD. But in this case, it was problematic. The nature of intelligence analysis is to gather as much information as possible to assist a policymaker in making difficult choices. If a policymaker has a preference for what the intelligence product should say, that pollutes the objectivity of the intelligence — and diminishes its value.
On Sunday, March 16, 2003, I watched Cheney on “Meet The Press” contradict our assessment publicly. “We know that he [Saddam] has a long-standing relationship with various terrorist groups,” Cheney said, “including the al-Qaeda organization.” I was basically watching Cheney field-test arguments that we would have to anticipate — and rebut — at CIA. Except instead of asking us questions behind closed doors, Cheney was asserting to the public as fact something that we found to be anything but. I found myself yelling at the TV like I was contesting a ref’s blown call in a football game.
The agency’s intelligence collection on Iraq’s relationship with al-Qaida was thin — Iraq’s connections to terrorist organizations were so minute it wasn’t a priority for us — so it was difficult to even construct a chart showing connections, as if we were mapping the Barksdale crew on The Wire. Saddam has a history of supporting small, anti-Israel terrorist groups; in early 2002, due to the war in Afghanistan, the terrorist leader Abu Mus’ab al-Zarqawi moved into Iraq on his own, with no direction or control by al-Qaida or Saddam; there were reports of varying reliability saying Iraq had discussions with al-Qaida about establishing a safe haven, dating from the early 1990s. The Zarqawi stuff would prove to be relevant, after the U.S. invasion. The rest of it didn’t add up to much. We concluded that, at most, the relationship between Iraq and al-Qaida was like two independent groups trying to exploit each other.
None of that stopped the invasion. Nor did the invasion stop the troll-y back-and-forth with the White House on Saddam and terrorism. When I volunteered to deploy to Iraq, my boss’ boss wouldn’t spare me for four days’ worth of weapons training. ( “I would rather have you come back in a body bag than spend that much time out of the office,” he told me.) They were so frantic to respond to White House questions that supporting the actual war effort took a back seat.
As it turned out, the questions wouldn’t stop once the invasion occurred. In June 2003, the Defense Department started to report that troops discovered caches of Iraqi Intelligence Service (IIS) documents allegedly proving Saddam was tight with al-Qaida.
The documents claimed to directly link Mohammed Atta, one of the main 9/11 hijackers, with the Baghdad training camp of Abu Nidal, the infamous Palestinian terrorist. It was a hand-written note, supposedly by Tahir Jalil Habbush al-Tikriti, the former head of the Iraqi Intelligence Service (IIS). If the documents were real, it was damning evidence that Iraq worked with al-Qaida long before the 9/11 attacks; if not, someone embarked on a very sophisticated strategy to play the U.S. government, our team or both.
And if this was truly a smoking gun, our team would be blamed — rightly — for getting it wrong. There was a sense of urgency at the CIA to evaluate these documents and provide an answer.
We got to work. We worked with the Secret Service to test the ink. If we could determine how old the ink itself was, we would have a time frame for when the documents were prepared. While in Iraq, I asked every high-level Iraqi government detainee I could about the details of the document. They either had grave doubts, or flatly said the document was bogus. They were adamant the names and roles outlined in the documents didn’t match the structure of the IIS. Having studied the Iraqi intelligence apparatus extensively before the war, everything they said lined up with the structure we understood to be true. The FBI had put together a timeline of Atta’s 2001 travels around the world, culling together airline records, ATM withdrawals and hotel receipts. Much like the infamous “Prague meeting” — another ultimately-bogus thread weaving al-Qaida together with Saddam — the FBI material indicated Atta was in the U.S. when the IIS document indicated he was meeting with Abu Nidal in Iraq.
Our Branch Chief, Karen, walked into Cheney’s office with everything we’d uncovered about the Abu Nidal link in June 2003. It seemed airtight. The Secret Service had determined that the paper was made after the date printed on the page. The timeframes didn’t match. The ink was inconsistent with the ink manufactured in the early 1990s purported timeframe of the documents. The chain of command indicated in the documents contradicted the description of the Iraqi intelligence bureaucracy provided by our detainees, even down to incorrect titles. These were forgeries.
I wasn’t there, but I heard the vice president was gracious and thanked her.
I actually quit the CIA for 3 days in 2004. I was exhausted answering historical questions trying to justify the invasion while at the same time trying to define Zarqawi’s growing role as a real threat. I couldn’t take it. People were dying and we were still talking about evidence of a connection between Saddam and al-Qaida. After a few phone calls with leadership in the Counterterrorism Center, I went back after 3 days and switched roles to the operations side — the National Clandestine Service — heading up the targeting operations team looking for Zarqawi. Instead of writing about him, I wanted to find him, I felt like the U.S. accidentally gave him a platform that helped him grow into a major terrorist. I moved onto another assignment a few months before he was killed in 2006.
After leaving the CIA, I’ve had a lot of time to reflect on this sorry absurd role in intelligence history, and my bit role in it. No intelligence analyst should have to deal with policymakers delving into intelligence work. It sounds bureaucratic and boring, but the distinction matters: CIA doesn’t have a policy agenda, it seeks to inform those agendas. Politicians and appointees have ideas for shaping the world. Mingling the two is a recipe for self-delusion and, as we saw in Iraq, failure.
November 18, 2012 at 09:03PM via Pocket
November 18, 2012 at 11:15AM via Pocket
For the third election cycle in a row, Nate Silver is winning praise as the guru of election forecasting. There’s a lot of confusion about what he does and doesn’t do, so here’s a quick guide to why his record is a big win for mathematics rather than gut instinct.
Who is Nate Silver?
Silver is a former economic consultant who developed a system for analyzing baseball player performance and forecasting the likelihood of their future success. He later adapted his techniques to political forecasting and won attention in 2008 when his forecast likelihoods came off in reality for 49 of the 50 states in the presidential election and all 35 Senate races. He later began writing for the New York Times and made predictions for the 2010 mid-terms, coming close but not quite as accurate as in 2008.
How did he do this time round?
In all 49 states (plus DC) where a Presidential race winner already looks clear, the candidate Silver ranked more likely to win did so (pictured). He had Obama a slightly favorite in Florida, which at the time of writing was not yet a clear result though Obama was ahead on votes that had been counted. Silver rated Obama a 90.9 percent favorite to win the Electoral College in his final analysis on election day. He also forecast a 50.8 to 48.3 percent popular vote lead for Obama; at the time of writing, the Associated Press count has Obama leading 50.4 percent to 48.0 percent.
So Silver correctly predicted every state and the overall race?
Not as such. Silver’s analysis is not intended as a prediction, but rather an assessment of the likelihood of winning. By definition there’s no way to confirm that assessment was correct as the real voting only takes place once: you’d have to run the race numerous times to even begin checking a likelihood forecast was correct. In the same way that a Romney victory wouldn’t have proven Silver wrong, an Obama victory doesn’t prove him right. What we can say is that taking all the individual states into account, his analysis certainly comes across more credible in hindsight than many rivals.
So how does he do it?
Silva’s talked about the details of his technique, but the explanation here is a slight simplification to show the general principles. It’s a two-stage process: first assessing the likelihoods in each state, than translating that into a nationwide prediction.
The main data gathering and analysis is simply gathering together all available opinion poll results, the logic being that polling is generally accurate but with statistical margins of error that should be reduced heavily by combining multiple polls. However, Silver doesn’t simply average the polls but rather weights them based on a combination of their sample size, how recently they were conducted, how they addressed the stated likelihood of respondents to actually vote, and the past performance of the polling company (a comparison of their previous polls and the actual results of the race concerned.) The idea is to minimize the effects of systemic bias in which the way a particular poll is conducted consistently affects the results in the same direction.
Does Silver do anything else with the numbers?
With all the numbers Silver crunches he applies non-polling data, though the way he does so can be misunderstood. The non-polling data involved factors such as the state of the economy, the benefit of incumbency, the number of registered party supporters in particular states and so on. Silver only applies this data where analysis shows a 90 percent certainty that it correlated to previous elections and, importantly, only uses such data to plug the gaps where recent polling data is insufficient. The non-polling data doesn’t ever replace polling data.
How does this translate into a national race?
While there’s some account taken of national polling and non-polling data, Silver primarily works through simulations using a system similar to the Monte Carlo method used in techniques such as weather forecasting. This involves running simulations of the state voting based on the forecast likelihood in each state.
To give a hypothetical (and extremely simplified) example, you could start with Oregon having a 90 percent chance of going for Obama, pick a random number from 0 to 100, and chalk the state up as an Obama win if the number is from 0 to 90. Repeat this process with each state and you get a national result and a resulting electoral college winner.
What you then do is run the process many many times over and track how many times particular outcomes and Electoral College totals come out. The idea is to get an idea of how likely it is to get different combinations of states going for or against the candidate predicted as more likely to win. This also helps measure the impact of the fact that, for instance, being wrong in Ohio is more likely to make the overall result different than being wrong in New Hampshire.
How did Silver’s forecasting differ to that of pollsters and pundits?
During the entire campaign, Silver always had Obama down as more likely to win overall. At his lowest point (just after the first TV debate) Obama was still rated a 61.8 percent likely winner, a likelihood that improved rapidly in the run up to Election Day when it became less likely voters would change their minds between responding to a poll and casting a ballot. Many pundits described the race as a toss-up or neck-and-neck right up to election day, refusing to label one side as more likely to win.
So why were the pundits getting such a different outcome?
There are three main reasons, the first two of them behavioral. Pundits may have been cherrypicking polls that met their preferred results and thus getting too excited about reading supposed momentum that may simply have been down to the natural statistical variation in polling. Secondly, some media outlets may have deliberately put an emphasis on anything that suggested an unpredictable race, whether to make a simple narrative that portrayed events as significant in affecting the outcome, or simply to make for a more interesting story than telling people the result wasn’t in doubt and risking them switching off.
The main issue however seems to be people misunderstanding what the numbers Silver produced are meant to portray. A forecast that, for example, Obama has a 75 percent chance of winning does not mean you are saying that he’ll win 75 percent of the vote or even that the final vote count in a state or national race won’t be close.
The forecast is a measure of certainty of the result, not a measure of the likely margin. In other words, Silver was saying that although the result might be close, it was highly likely Obama would win — something that’s partly the result of the Electoral College state breakdown working in his favor.
In fact by the time of his forecasts in the past few days, Silver argued that an Obama win was virtually certain based on the mass of polling — the chance he gave Romney was effectively nothing more than an assessment of the possibility that the polls as a whole were systemically biased.
Ah, but Silver got the electoral college total wrong didn’t he?
Not as such. Silver did give a figure of 313 for Obama’s electoral college total but again this wasn’t a prediction, rather an average of all the simulated outcomes. This doesn’t translate to a forecast as 313 may not even be a possible total, and several previous daily assessments that had fractions clearly weren’t.
Simply as an average, the figure has come out well of course: it lies between the two possible totals (303 and 332) that are left with Florida yet to be confirmed. Whatever the result, it’s within the margin created by the closest state.
What’s more impressive though is that Silver also produced a chart of how many times each individual total appeared in his simulations. Top place, coming up as a 20 percent probability, was 332. Second place, with 16 percent, looks to be right on 303.
No related posts.
October 28, 2012 at 02:56AM via Pocket
November 06, 2012 at 06:59AM via Pocket
October 26, 2012 at 02:51PM via Pocket
October 24, 2012 at 12:43PM via Pocket
October 14, 2012 at 12:37AM via Pocket
October 20, 2012 at 02:45PM via Pocket