Support “Altered Perceptions” #

Rob is a friend of mine. Or, maybe more accurately, was a friend, because I’m awful at correspondence and keeping in touch with people. He’s a smart guy, and think highly of him. I remember sitting in the lounge at school, between classes, and just shooting the breeze with Rob. He was the only student in the cohort that was a published author. He’s easy to talk to, and an interesting person as well.

Rob and his friends have put together a fantasy anthology of alternate and deleted scenes from some well-known (and some new-to-me) fantasy authors, with the proceeds set to help Rob get out of some debt that’s been exacerbated by mental health issues, and to raise awareness and funding for mental health in general. The list of authors donating their work is impressive, and it’s worth supporting. I did.

I wish I understood mental illness better. It’s all too easy to marginalize those dealing with it. I’m glad of this project, because it makes it more real to me.

Rob and his family were among the many who helped us out when we were dealing with my son’s cancer. This time it’s his turn, so go donate, and get a great fantasy anthology at the same time.

Slate: America’s schools are segregating again #

In recognition of the anniversary of Brown v. Board of Education, Slate concluded schools are segregating again. Not really a surprise, given the ethnic maps I linked to previously—my anecdotal experience is consistent with the conclusion. My son, for example, attend a majority black school. Last year, he was the only white kid in his class. Our district is predominantly white; the school district across the street (on three sides) is predominantly white.

The average white student, for instance, attends a school that’s 73 percent white, 8 percent black, 12 percent Latino, and 4 percent Asian-American. By contrast, the average black student attends a school that’s 49 percent black, 17 percent Latino, 4 percent Asian-American, and 28 percent white. And the average Latino student attends a school that’s 57 percent Latino, 11 percent black, 25 percent white, and 5 percent Asian-American.Jamelle Bouie, Brown v. Board of Education 60th anniversary: America’s schools are segregating again., Slate, 15 May 2014

Interesting. Not sure about the methodology: In some of the states out West, schools are predominantly white because the population of the state/county/city is predominantly white, so this would tend to greatly skew the results of white participants. I did start to see some schools in Utah with large Hispanic/Latino populations when other schools in the same district had much lower percentages. I wonder whether blacks are more likely to congregate than other cultures.

The key point, however, is this:

School segregation doesn’t happen by accident; it flows inexorably from housing segregation. If most black Americans live near other blacks and in a level of neighborhood poverty unseen by the vast majority of white Americans, then in the same way, their children attend schools that are poorer and more segregated than anything experienced by their white peers.Ibid.

I have visited twenty or so public and private high schools this year. The differences between those in the richest and poorest neighborhood are near appalling. As previously linked, states can get involved by implementing policies that bring greater funding equity across districts. There is also a social gap in college financial aid. Surely there are solutions that can be implemented on the housing side as well.

Some of the difficulty is that race is still too strongly correlated with income. We haven’t yet overcome generational effects.

“Income has become a much stronger predictor of how well kids do in school,” Reardon says. “Race is about as good a predictor as it was 30 years ago. It’s more that income has gotten more important, not that race has gotten less important.”

Sarah Garland, “When Class Became More Important to a Childs Education Than Race”, The Atlantic, 28 August 2013

Effectiveness of State Education Funding Equalization #

From, a summary of some work by the National Bureau of Economic Research.

Title: “The Effect of School Finance Reforms on the Distribution of Spending, Academic Achievement, and Adult Outcomes”

Authors: C. Kirabo Jackson, Rucker Johnson, Claudia Persico

What they found: School finance-reform efforts have led to more equal funding for education, which has, in turn, helped students from poor families stay in school longer and earn more in adulthood.

Why it matters: Because most of the country funds public education through local property taxes, school districts in affluent areas have historically spent far more on a per-student basis than ones in lower-income areas. Various programs have aimed to address the issue, but it hasn’t been clear how successful those efforts have been in either reducing inequality or improving student outcomes. The authors use newly released spending data to conclude that the programs have indeed reduced inequality in funding, and that court-ordered reforms have been more effective than legislative ones. They also find that increases in spending lead to higher graduation rates among students from poor families, as well as higher earnings and reduced poverty when those students reach adulthood. They find no impact for students from nonpoor families.

Key quote: “These results provide compelling evidence that the [school finance reforms] of the 1970s through 2000s had important effects on the distribution of school spending and the subsequent socioeconomic well-being of affected students. Importantly, the results also speak to the broader question of whether money matters. … Many have questioned whether increased school spending can really help improve the educational and lifetime outcomes of children from disadvantaged backgrounds. The results in this paper demonstrate that it can.”Data they used: Historical Database on Individual Government Finances, the Local Education Agency School District Finance Survey and the Panel Study of Income Dynamics, among other sources

Ben Casselman, “In the Papers: A Look at the First Major Government-Sponsored Welfare Program”,, 12 May 2014.

Regular Expression Crossword Puzzle #

Via Kottke and @grimmelm, the Internet delivers a wonderful regex crossword.

Difficult, entertaining, and, as @grimmelm says, “It’s not as bad as it looks.”

I did it in pen. Successfully.

Still, through the middle third of it I kept wondering if it would be easier to write a back-tracking search tree program to solve it for me…

Yes, there is a single, unique solution.

TSA Confiscates “Dangerous” Stuffed Toy Accessory #

A TSA agent in St. Louis made air passengers everywhere rest easier by confiscating a “dangerous” two-inch prop gun that was accessorizing a stuffed animal. [via Lowering the Bar]

On Monday, the TSA issued a statement, saying “TSA officers are dedicated to keeping the nation’s transportation security systems safe and secure for the traveling public. Under longstanding aircraft security policy, and out of an abundance of caution, realistic replicas of firearms are prohibited in carry-on bags.”Susan Wyatt, “TSA agent confiscates sock monkey’s pistol”, KING 5 News, 8 December 2013

Right. Because a 1/5-scale toy the size of a couple of quarters is a “realistic replica.”

Muppets: Swedish Chef and Gordon Ramsey #

A Sociologist Interrogates the Criminal-Justice System #

Great story from the Chronicle of Higher Education about a sociologist embedded in a poor, high crime neighborhood.

Via Next Draft.

iOS 7.0.4 Update Caused Data Loss #

I’m not having much luck with software updates this season.

I happily clicked “Agree” to update my (not jailbroken) iPhone 5 to the most recent software update, iOS 7.0.4. Something must have gone wrong, because it dropped to an error screen that insisted the phone by plugged in to iTunes … and the only this iTunes would do with it was a full restore.

My last phone backup was a month ago. One month of data, pictures, call logs, save games: gone.


Jenkins CI Install Failure on OS X #

I’m sampling Continuous Integration (CI) tools for a project I’m working on. One of the most ubiquitous open source options in Jenkins, which comes with a convenient package installer for OS X.

It installed without errors, but when it came time to run Jenkins (browsing to http://localhost:8080/ ), my browser(s) wouldn’t connect.

The installer built its own log files as \var\log\jenkins\jenkins.log, which helped unravel the mystery: Java wasn’t installed.

Huh? I’ve taught courses in Java from this machine. Java was installed.

Turns out, upgrading to Mavericks “helpfully” removed Java without telling me. A placeholder app is still there in /usr/bin/java, but it simply loads an alert prompt to download and install Java … an alert prompt the fails silently when run by a daemon (which by definition can’t access windowing functions in the OS), like Jenkins.

As LaunchDaemon will attempt to re-run the failing Jenkins every 10 seconds, turn it off temporarily if you need to (re)install Java:

sudo launchctl unload -w /Library/LaunchDaemons/org.jenkins-ci.plist

Re-enable by repeating the same command, but using load instead of unload.

CMU Password Cracking Study #

The landmark study is among the first to analyze the plaintext passwords that a sizable population of users choose to safeguard high-value accounts. The researchers examined the passwords of 25,000 faculty, staff, and students at Carnegie Mellon University used to access grades, e-mail, financial transcripts, and other sensitive data. The researchers then analyzed how guessable the passwords would be during an offline attack, such as those done after hackers break into a website and steal its database of cryptographically hashed login credentials. By subjecting the CMU passwords to a cracking algorithm with a complex password policy, the researchers found striking differences in the quality of the passwords chosen by various subgroups within the university population.Dan Goodin, “It’s official: Computer scientists pick stronger passwords”, Ars Technica, 8 November 2013.

One of the funnier conclusions: Those associated with the business school tended to have the weakest passwords.

A very unusual data set, available due to remarkable circumstances:

Plaintext passwords were made indirectly available to us through fortunate circumstances, which may not be reproducible in the fu- ture. The university was using a legacy credential management system (since abandoned), which, to meet certain functional re- quirements, reversibly encrypted user passwords, rather than using salted, hashed records. Researchers were never given access to the decryption key. Mazurek, et al. “Measuring Password Guessability for an Entire University” [pdf], 22 October 2013.

From reading the paper, the “cracking” was based on guessing from pre-composed password lists, based on publicly leaked lists, and experiments with Mechanical Turk.

Super interesting. The steps researchers had to go through to protect privacy and keep the IRB happy are exceptionally thorough, including code review and secure facilities.

We were required to submit all the analysis software needed to parse, aggregate, and analyze data from the various data sources for rigorous code review. Upon approval, the code was transferred to a physically and digitally isolated computer accessible only to trusted members of the university’s information security team. Through- out the process, users were identified only by a cryptographic hash of the user ID, created with a secret salt known only to one infor- mation technology manager.

We were able to consult remotely and sanity-check limited output, but we were never given direct access to passwords or their guess numbers. We did not have access to the machine on which the passwords resided — information security personnel ran code on our behalf. Decrypted plaintext passwords were never stored in non-volatile memory at any point in the process, and the swap file on the target machine was disabled. All analysis results were personally reviewed by the director of information security to ensure they contained no private data. We received only the results of aggregate analyses, and no information specific to single accounts. After final analysis, the source data was securely destroyed.


Hire Tom! Hire Tom!