* Posts by Archimedes_Circle

61 posts • joined 3 Mar 2011


Software bug sets free thousands of US prisoners too early


Perhaps because the number of days that someone is released early from prison, amongst people who were released early, is a Poisson process, which means that it is discrete, and could quite likely be non-symmetric. So they chose a value that better indicated the centre of the distribution and was less prone to outliers.

Why use any number?

They're all just somebody's lies.

Halo 5: Overhyped, but still way above your average shooter


Re: Rating

...most games nestle on a 7 for me because they're above average sure, but by no means amazing.

How can most of anything be above average? I'm not trying to be overly picky, but I think this is part of the problem. People feel bad when they give a five, so they give a seven instead, and that becomes the new `average'. At the same time, the consuming public comes and thinks, `ooh, it's above average, maybe I'll pop out and purchase this wonderful experience of pomp and circumstance,'' when in actuality it really is just average.

What we need to do is recentre the distribution after everyone has given scores, and see how above or below average a game is, using the average for each site to centre new scores, just like z tests work, if you remember your statistics. Then aggregate across the z-scores and you'll have a good determination of whether a game is, for the average website, rated as above or below their average rankings.



Re: Who killed them?

Can entropy ever be reversed?

Computers know you better than your friends


Re: Well I've shared this on facebook

The definition of the statistical procedures they're doing here (factor analysis, so they're asking if the regression slopes are statistically different from 0 between the items and a shared commonality between the items, called the latent factor[s]), combined with the fact that they're publishing, means that these results are predictive. The personality factors measured here aren't quite `likes open spaces, but only before 17:00' but the Big 5: openness, conscientiousness, extraversion, agreeableness and neuroticism, and scores on them, as calculated from covariances between the 100 observed items in the test itself, do predict a fair amount. Lots of gene and personality links, authoritarianism (Bob Altemeyer's right wing authoritarianism/ Jost's cognitive flexibility). Not predictive to the extent that HR consultants sell them as though, and my personal past time has been quizzing them on how they work. Blank stares are pathetically common, and they just don't understand that just because they establish a test amongst grocery bag packers, doesn't mean that the same qualities are universally beneficial for statisticians, or sysadmins.

As for whether anything that can be meaningfully predicted is either useful or relevant is more of a judegement call.

What can The Simpsons teach us about stats algorithms? Glad you asked...


Re: So... what's the options, then?

You could do multilevel modelling as a solution. Multilevel modelling is, in general, a solution to when subjects are clustered, like when you want to explore the effects of spending on student performance. Typically you would run a regression, however in this case, if you have multiple schools, the spending levels are not independent observations: they are correlated with each school and that typically inflates the wald test results, which makes smaller p-values.

Code-busters lift RSA keys simply by listening to the noises a computer makes



I think it is worth pointing out that while 2.x does include blinding schemes to avoid side channel leakage, as of http://lists.gnupg.org/pipermail/gnupg-announce/2013q4/000337.html so does gpg 1.4.16, which was added following private contact by one of the researchers.

As for those wondering why it was left out, previously local LAN attacks allowed sidechannel leakage to exploit this but there was no other vector known until this paper. However gpg 1.x isn't designed for such purposes therefore it was left out. Doesn't mean someone did not deploy it incorrectly, but gpg has always been rather conservative.

Gaming co ESEA hit by $1m fine for hidden Bitcoin mining enslaver


Re: Pipe(lined?) dream

FGPA or ASIC hardware seems to be the best bet.

TrueCrypt audit project founder: 'We've set our sights high'


Re: Still a problem for non-techies


1) A perfect one-time pad gives perfect and unbreakable security.

2) Key sizes are ridiculously short for no reason. They should be in MB, not KB for anything important

3) PKI is badly broken, Key sizes are too small, algorithms are suspect, there are no trustworthy CAs, etc, etc.

4) The weakest link in even well designed systems is the entropy source used for the generation of keys, nonces, salts, etc. </blockquote>

1. OTP is the only perfect security. However, key management is an unsolvable problem as of yet. Once I've exhausted the one terabit key file I sent you, we need to re-exchange again, with no efficient way other than trusted courier. Furthermore, the benefit of asymmetric encryption is that I don't need to know you before hand: All I need to have is your public key. However, OTP (and any other pure symmetric encryption process) all fail the bootstrapping problem with respect to trust and key exchange. With respect to other symmetric options, we would need to exchange a password before we could ever communicate electronically.

2. MB (or Mb) sized keys would be ridiculously inefficient, and provide no improved security over the standard 256 bits of security we want now. Focusing first on asymmetric encryption: RSA's security to efficiency ratio peaks at 3072 bit key size, which is ~115 bits of security. After that, the gains are minimal compared to the massive increase in the size of the key. Elliptic curve cryptography is the next stop, with a much cleaner conversions between asymmetric key size and symmetric key security, 512 bit ECC = 256 bit symmetric.

SInce encryption strength is typically evaluated in terms of symmetric keys, we can now assume that all complexities are functions of symmetric bit size. Now we get into physical limits of the universe, and something called Landauer's principle (this is an excellent overview of the details: http://security.stackexchange.com/questions/6141/amount-of-simple-operations-that-is-safely-out-of-reach-for-all-humanity). Basically though, it states that 128 bits of security will be broken in 2040, which practically translates into 2050 being the year your key is broken, given that we consume the entire planet's energy resources, which was consumed in a decade, starting in 2040. There's some unrealistic assumptions in there that make this an unrealistic best guess with respect to the timeframe.

Now, the other concern to this is that there is an efficient algorithmic break that drastically reduces the key space to evaluate. Of course, if this is true, then any size key using the same algorithm, will be susceptible, and thus no gain.

4. Technically, salts do not need to be random, or even unique. They just are appended to existing passphrasses avoid rainbow table cracking. These are no longer an acceptable practice, thanks to GPU hashing. Much better would be to utilise something like bcrypt, scrypt or PBKDF2, which are not designed to be computationally cheap. That said, I agree that entropy is a failure point, and we need multiple independent sources, mixed together, to counteract suspicions like those about Intel's chip flaws.

Fear the Embarrassing Bodies webcam


Re: House!

He ended up in America, playing Gregory House. Also did "A bit of 'fry and laurie'" I believe.

Kinky? You're mentally healthier than 'vanilla' bonkers


Re: Tells us more

What you're largely referring to is clincal psychology, a field that doesn't have any claim to science, and which has a pretty active history of ignoring glaring flaws in favour of their ad hoc 'expertise'. Look up the 1954 book by Paul Meehl, or the 2000 meta-analysis which confirms that. Unlike the rest of the field, clinical psychology hasn't really moved on beyond Freud


Re: WTF?

<quote>"Associations were examined using χ2 tests of independence with φ and Cramer's V as effect size measures and eta or Pearson's correlation. Group differences were tested using analysis of covariance, with partial η2 as effect size measure. A priori contrasts were tested using α = 0.01 to correct for multiple testing; for all other tests we used α = 0.05, two tailed."</quote>

Effect size measures are the expected differences you would find, given the p values expressed. In other words, it's a projected value of what the 'effect' of the difference would be. Phi and Cramer's V are correlations between nominal/categorical values, since pearon's correlations doesn't work for things like categorical crosstabs. Without getting the paper, they did something like numbers of males/femals who have/haven't been admitted to a psychiatric ward, or any other form of group membership. ANCOVA(analysis of covariance) is just t tables for multiple groups, with some expected differences controlled for (removed). For example, there may be systemic differences between different genders at two different college campuses, and I want to pose the question: do women graduate more often then men do, regardless of family income levels. You control for SES (previously measured) and find out if there are still differences, or if the gender differences are actually the result of SES differences.

The alpha level choice is a bit weird, because the language seems to indicate that they chose it regardless of the tests they ran, but it would have been much easier to use either the Bonferroni or Sidak methods of correcting. Essentially, when you do multiple group testing, you can accidentally deflate the p values, and receive significant differences where non exist.

Two tailed just means they looked at confidence intervals around the top and bottom of the distribution, rather than assigning the CI to one side or the other; standard practice, since the distribution is assumed to be gaussian. Oh, and the partial bit just means the correlations after controlling for differences.

/research psychologist, although one who doesn't find this area of research to be anything useful.

Google to double encryption key lengths for SSL certs by year's end


Re: What do they know?

You would need 1024 qubits to factor a 2**1024 coprime integer. I thought D-Wave was only a handful. If there was a security compromise do you really think d wave would still be on business to anyone other than the NSA?

Report: BlackBerry BYOD-ware doesn't pass UK.gov security test



K-9 and APG to provide gpg signing and encryption

Textsecure for encrypted texting records and messaging, although why doctors need to discuss patients over texting is not something I understand.

Gibberbot for jabber/gmail encrypted chat comms.

Redphone for voice.

And android built in encryption. I'm not going to claim android is better than IOS for full disk encryption, but if that's your threat model the game is already over, because you shouldn't have copies of secured docs sitting on your phone.

Google files patent for eyewear that SHOOTS LASERS


Re: Was excited because I misread the headline...

Thats the lawyer gom jabbor test. If they file the patent for the device, they fail the self preservation test.

Actually, would filing it prevent cheap knock offs, actually reducing the supply available to the public?

British armed forces get first new pistol since World War II


Re: The point is not always to kill

Robocop actually has a heavily modified Beretta 93R (with a 50 round clip :P). The character played by Kurtwood Smith though had numerous versions of the Desert Eagle.

Android apps get SSL wrong, expose personal data


iOS apps

From my interpretation of this, as well as another colleague's, this is really a problem with not having proper public key pinning. This isn't limited to Android, it's just that the researchers chose to focus on Android.

Ay caramba, Ubuntu 12.10: Get it right on Amazon!

Big Brother


sudo apt-get remove unity-lens-shopping. Sorted, though I haven't decided to take the plunge to Ubuntu 12.10

Drinking too much coffee can MAKE YOU BLIND


Re: In other news

Please, that's nothing. Oxygem is no addictive that going cold turkey kills you in minutes.

Barnes & Noble Nook HD and HD+ hands-on review


Re: re. multiple user profiles

You know, the average bodice ripper has the equivalent of a us 3rd grade reading level of difficulty.So actually, the profile is more accurate than you'd think.

Health minister warns ISPs: Block suicide websites or face regulation


Re: Actually...

Look up social proof and the woether effect (aka copycat suicide). I'm not advocating censorship, but there is a huge correlation between media coverage and suicide rate upticks. In the context of social proof, people identify with other people and are then more likely to copy their behaviour (no shit right?). Anyway, what group is more likely to be identified with than the PR managed image of celebrities, who are designed to resonate with target demos. One offs themselves and a huge chain of fans follow. This deserves better, but I'm typing on a phone, so I'll call it an end. Again, not calling for censorship, but a little less media sensationalism.

Ten netbooks

Thumb Up

I find it a drop in replacement, ymmv. Better choice would be dualboot and just mount the ntfs partition and run everything. Worse case you have to delete the partition afterwards.

Sony pushes patent for interactive TV ads


Re: ITVPlayer...

Hulu already implemented this exact premise with that DS puzzle game, Dr something or other. There was an interactive ad where you had to finish the maze in x seconds or wait 30 or 45 seconds for it to finish.

Don't bother with Big Data – listen to customers instead


Observations instead of data?

Since when are observations not data then? If "data" needs to be replaced, then you're not observing the right data.

Smutty books strip Harry Potter of Amazon crown

Paris Hilton

Re: They're shite.

The reminds me of an economics study published back in the spring. It looked at the Oprah Book Club sales and found some interesting conclusions. First and foremost, significantly less books were sold because of Oprah. It turns out that those who watch Oprah read at an average level of 3rd grade, coincidentally what the craze of bodice rippers have been for the last forever or so. It also turned out that the Oprah books were significantly longer than the usual tripe and three grade levels harder (6th). So what they ended up with was longer and harder (COMPARATIVELY mind you. I agree 6th grade reading isn't exactly something worth breaking out the OED over) books that consequently meant less books were read. Worth a read,

Paris....well because the parent's comments on boring sex and thin are fitting.

Foreign intelligence agencies are biggest online threat, ex-Fed warns


Misread that as French, not foreign. That was no hell of a double take.

Crypto boffins: RSA tokens can be cracked in 13 MINUTES


Re: May be a dumb question but

The amount of security you gain by increasing the key size decreases rather quickly, especially when performance is factored in. Or at least that's the traditional model/assumption. There was an interesting thread on the openGPG mailing list last month, subject ="Some people say longer keys are silly. I think they should be supported by gpg."

The OP was a nut, but it did result in some useful chatter. The main issue though is that under powered(mobile) hardware can't handle huge keysizes without creating an equally huge latency. That said, I run 4096 RSA keys on my phone without an depreciable lag, but many of the older OpenGPG members disagree.

However, NSA current guidelines establish that once you go beyond 4096 bit security (actually I think it's 3072 bit) a better option is to switch to Elliptic Curve Crypto. That is far more efficient in terms of size. Normally security is scaled in X bits of symmetric cipher, and ECC 512 bit is = to 256 bit security; ECC-256 is 128 bit security. On the other hand, RSA 4096 is somewhat like 142 bits. Doubling that to RSA 8192 only ups the security to 194 bits. That's a huge increase in keysize (overhead) for very little security. So it's half that nobody will ever need more than 64 kb of ram, and the rest is that you cannot predict a break in a cipher system that means your security is not worthwhile, and the lag you introduced may present timing attacks, and useless overhead.

Something like TWIRL cuts off 11 bits of security off of anything involving number factoring, and obviously quantum computers would shred through any RSA cipher.

Finally a use for quantum computers: Finding LOL-cats faster


Re: Don't panic!

Thank you, saved me a rant. However, symmetric key sizes are effectively halved, so that 2**256 retains the effective strength of 2**128. Also, ECC is vulnerable to an adapted version of shor's algorithm. There are several QC secure algorithms, with homomorphic encryption being first off my mind.

Flying Dutchman creates dead cat quadcopter


Re: Flying pigs next?

Hey, Swine Flu

World+Dog to demand ever larger tablet-phones



What with all this talk of ladies and miss's hands, I shan't think this is a geek website anymore. Now, back to your basements! The DM is getting lonely.

Newly discovered asteroid will not ANNIHILATE THE EARTH


Eh, so anyone try to figure out whose satellites it's going to smack into? It's 16,000 miles out from the center of the earth. I declare Mardi Gras all week just in case it hits something important, like a black budget viral storage lab that lands earth side a la Andromeda Strain.

Finders of lost mobes can't resist staring at privates


Couple of things

There is, admittedly, a smidge of alcohol behind this overview, but anyway:

1. Yes,the percentages make no sense whatsoever.

2. I desperately wish to look at the indivudual data for the latency between discovery of phone and access attempt. x_bar=10.2 but the median is 59 minutes? Who waited forever? Stnd. deviation would be far more useful.

3. Point number two leads into this one: the data supposedly dropped off for all but a few after seven days.....which smartphone lasted for 7 freakin days? Seriously. I can get maybe 20 hours if I'm good on my phone. I guess they could have installed extended batteries, but still. 7 days with continuous or near continuous gps fixes?

4. Again, on that note, did any of the phones die before being accessed? At least two of them disappeared completely. Another 5% (again with the weird percentages) were moved but not accessed.

Wouldn't a more interesting study be leaving 50 phones with some sort of pin code enabled, and then look at people do with those. After all, what IT policy would allow a phone to not be secured and encrypted. Look at how many people are actually interested in getting into these devices when the bar is set a little higher. As opposed to empiracle evidence that human curiousity is a bad thing for secrets. Pandora's Box, Garden of Eden parables anyone?

Interesting though, the project head's original research was in looking at the same concept, dealing with usb sticks. Stuxnet anyone? Or the blinkin obvious.

Heyhey! Maybe the annoying percentages are due to the fact that 47 devices were reported upon in some cases, not 50. No, guess not.

And what is this about "Note that on most smartphones, there is not an easily accessible file system as there is on personal computers. So, document type files less commonly appear on the home screens of most smartphones. Therefore, the majority of the icons on the devices leveraged for this study represented apps that initiate a program on the device"

Each version of android I've played with has had a file manager, and it's not like it's difficult to access adobe and have it pull up accessed files. Who honestly thinks someone keeps HR-UBERSECRET.pdf on their homescreen.

Recommendations: Do not lose your phone....NO SHIT Sherlock.

Google Wallet PIN security cracked in seconds


Salting is good, but where would you keep the salt? Either it would be encrypted on the phone (and thus either having the decryption key on the phone, or the person keys it in which would make it easier to just have a longer password.) A network connection is tricky because losing a connection leaves you out, plus it's rather roundabout. And you're screwed if the server dies.

Try a 'shroom before ruling on chill pills, boffin tells gov


Thank God we still have homosexuality on the books too. Wouldn't want those queers feeling normal. Why can't they handle the truth too.

Course, I'm still waiting for when the pedophiles march into the schools like the pied piper because we let the gays marry. Seems like MAYBE just because something is illegal doesn't necessarily imply it's immoral.

iPad users 'risk shoulder pain', say US gov, Microsoft boffins


Apparently there's more than we thought!

Clearly a (E)UNIX guy.

YouTube tops one hour of video uploads per second


For those on the right side of the pond's persuasion I apologize.

I can proudly (alright somewhat proudly) say that I prefer listening to nyan cat more than I prefer listening to several other types of music, especially those that refer to tractors and god with equal importance and frequency. I don't care how religious you are, or how proud you are of your damn green tractor. NO ONE thinks it is sexy.

/rant from years of pent up childhood misery

As for the title of the post, I've warned you. There be mental demons there, and Tim the Enchanter help me if you look.

The Register to publish other sites' blacked-out content in SOPA protest


What 9 letter word?

You can't mean a can opener. That's two words.

What are you on about?

Russian boffins: US radar didn't fry Phobos-Grunt

Black Helicopters

Looks like I'm out of a job

Anyone looking to employ an ex USG scientist who's working on a space probe destroying radar system?

Ten... laptop accessories


I think you'll find it's closer to 9% than 15%. 10% percent at the outside. I'd link to wikipedia but people keep telling me there is something wrong with the internet when I do.


US military access cards cracked by Chinese hackers


Technically, the Israelis designed that weapons system. Point taken though.

Cracked emails again deployed against climate researchers


The key word there is reduce. Your response is the equivalent of saying "I'm going to reduce my caloric intake," and then never eating again.

Vatican mulls God particle, calls for appointment of antichrist


You send us to Revelations for clarification AND information? For shame. Next you'll send us to baptists for advice on breweries.

This weekend: First ever iPADS IN SPAAAACE


How is it supposed to be entertaining? It can't even pr0.....I mean flash video.

Pay Jobs due respect - by crushing the empire he created


Must have been an interesting night.

scene: Mark Shuttleworth comes to the office late one night. Knocks on Assay's door and enters.

S: Mark! I've got some gre.........................MATT, WHAT THE HELL ARE YOU DOING WITH ONE OF THOSE.

A: Oh, ummm. ahhh..........market.......research?

S: You're not USING IT, are you?

scene fades, while it takes Richard Stallman time to travel to Canonical to take part in The Shunning Ritual.

On a side note: Has there been a single article written about Steve Jobs that did not directly quote, or at least reference, his Stanford speach?

Crims prefer old exploits: Microsoft


Is that what it said? Retroactive continuity! They make it much clearer now.

Thai prime minister Twitter hack suspect charged


Not a new word but technically the term is cracking, but we rarely make that distinction these days. As for the level of skill involved (or lack thereof)? An exploit is an exploit. If everyone actually used perfect password security, rainbow tables and dictionary attacks would be useless, and nothing would ever be compromised without a zero day exploit gone unpatched......in theory.

Unity: 'We'll make a terabit chip by 2014'


How else do you think people used to advance in leaps and bounds if not for people having not yet invented gravity?

Samsung and Google's next Nexus specs leak

Paris Hilton

That's what she said

The life and times of Steven Paul Jobs, Part One


Not to be morbid or anything, but how long have you had this one ready?

And more importantly, can I proof the copies of Linus' and Bill's? Or even Ballmer's?!

Edited for being rejected:Seriously, it's perfectly acceptable to have had this written in advance, considering his health complications, you wouldn't be performing due diligence if you didn't. Doesn't mean I can't poke the angry bear from time to time. I would still love to have Ballmer's reg edited biopic though.

Firefox devs mull dumping Java to stop BEAST attacks

Paris Hilton

Just installed no script again, after leaving it for some arbitrary reason I've long forgotten. Firefox literally flies. (and useless exposition is a joy)

Facebook: 'We don't track logged-out users'


>Benign names

So as long as it doesn't sound threatening, it can't be used to track you? By that logic, the Manhattan Project was somehow about Manhattan. Urban renewals or something I guess.



Biting the hand that feeds IT © 1998–2020