Advances in Social Network Analysis and Mining Conference — Sydney

This conference will be in Sydney in 2017, from 31st July to 3rd August.

http://asonam.cpsc.ucalgary.ca/2017/

As well as the main conference, there is also a workshop, FOSINT: Foundations of Open Source Intelligence, which may be of even more direct interest for readers of this blog.

Also I will be giving a tutorial on Adversarial Analytics as part of the conference.

Even more security theatre

I happened to visit a consulate to do some routine paperwork. Here’s the security process I encountered:

  1. Get identity checked from passport, details entered (laboriously) into online system.
  2. Cell phone locked away.
  3. Wanded by metal detection wand.
  4. Sent by secure elevator to another floor, to a waiting room with staff behind bullet-proof glass.

Here’s the thing: I got to carry my (unexamined) backpack with me through the whole process!

And what’s the threat from a cell phone in this context? Embarrassing pictures of the five year old posters on the wall of the waiting room?

I understand that government departments have difficulty separating serious from trivial risks, because if anything happened they would be blamed, regardless of how low-probability the risk was. But there’s no political reason not to make whatever precautions you decide to take actually helpful to reduce the perceived risks.

Secrets and authentication: lessons from the Yahoo hack

Authentication (I’m allowed to access something or do something) is based on some kind of secret. The standard framing of this is that there are three kinds of secrets:

  1. Something I have (like a device that generates 1-time keys)
  2. Something I am (like a voiceprint or a fingerprint), or
  3. Something I know (like a password).

There are problems with the first two mechanisms. Having something (a front door key) is the way we authenticate getting into our houses and offices, but it doesn’t transfer well to the digital space. Being something looks like it works better but suffers from the problem that, if the secret becomes widely known, there’s often no way to change the something (“we’ll be operating on your vocal cords to change your voice, Mr. Smith”). Which is why passwords tend to be the default authentication mechanism.

At first glance, passwords look pretty good. I have a secret, the password, and the system I’m authenticating with has another secret, the encrypted version of the password. Unfortunately, the system’s secret isn’t very secret because the encrypted version of my password is almost always transmitted in clear because of the prevalence of wifi. Getting from the system’s secret to mine is hard, which is supposed to prevent reverse engineering my secret from the system’s.

The problem is that the space of possible passwords is small enough that the easy mapping, from my secret to the system’s, can be tried for all strings of reasonable length. So brute force enables the reverse engineering that was supposed to be hard. Making passwords longer and more random helps, but only at the margin.

We could instead make the secret a function instead of a string. As the very simplest example, the system could present me with a few small integers, and my authentication would be based on knowing that I’m supposed to add the first two and subtract the third. My response to the system is the resulting value. Your secret might be to add the first and the third and ignore the second.

But limitations on what humans can compute on the fly means that the space of functions can’t actually be very large, so this doesn’t lead to a practical solution.

Some progress can be made by insisting that both I and the system must have different secrets. Then a hack of either the system or of me by phishing isn’t enough to gain access to the system. There are a huge number of secret sharing schemes of varying complexity. But for the simplest example, my secret is a binary string of length n, and the system’s secret is another binary string of length n. We exchange encrypted versions of our strings, and the system authenticates me if the exclusive-or of its string and mine has a particular pattern. Usefully, I can also find out if the system is genuine by carrying out my own check. This particular pattern is (sort of) a third secret, but one that neither of us have to communicate and so is easier to protect.

This system can be broken, but it requires a brute force attack on the encrypted version of my secret, the encrypted version of the system’s secret, and then working out what function is applied to merge the two secrets (xor here, but it could be something much more complex). And that still doesn’t get access to the third secret.

Passwords are the dinosaurs of the internet age; secret sharing is a reasonable approach for the short to medium term, but (as I’ve argued here before) computing in compromised environments is still the best hope for the longer term.

Security theatre lives

Sydney tests its emergency notification system in the downtown core at the same time of day every time. So if a person wanted to cause an incident, guess what time they would choose?

It also seems to be done on Fridays, which is exactly the worst day to choose, since it’s the most common day for islamist incidents.

Security theatre = doing things that sound like they improve security without actually improving them (and sometimes making them worse).

“But I don’t have anything to hide” Part III

I haven’t been able to verify it, but Marc Goodman mentions (in an interview with Tim Ferriss) that the Mumbai terrorists searched the online records of hostages when they were deciding who to kill. Another reason not to be profligate about what you post on social media.

The growing role of data curation

My view of Data Science, or Big Data if you prefer, is that it divides naturally into three different subfields:

  1. Data curation, which involves focusing on the issues of managing large amounts of heterogeneous data, but is primarily concerned about provenance, that is tracking the metadata about the data.
  2. Computational science, which builds models of the real-world inside computer systems to study their properties.
  3. Analytics, which infers the properties of systems based on data about them.

I’ve posted about these ideas previously (https://skillicorn.wordpress.com/2015/05/09/why-data-science/),

Data curation might have seemed like the poor cousin among these three, and certainly gets the least funding and attention.

But issues of provenance have suddenly become mainstream as everyone on the web struggles to figure out what to do about fake news stories. So far, the Internet has not really addressed the issues of metadata. Most of the big content providers know who generated the content that they create and distribute, but they don’t necessarily make this information known or available for those who read the content to leverage. It’s time for the data curation experts, who tend to come from information systems and library science, to step up.

Data curation is also about to become the front line in cyberattack. As I’ve suggested (Skillicorn, DB, Leuprecht, C, and Tait, V. 2016. Beyond the Castle Model of Cybersecurity.  Government Information Quarterly.), a natural cyberdefence strategy is replication. Data exfiltration is made much more difficult if there many, superficially similar, versions of any document or data that might be a target. However, progress in assigning provenance becomes the cyberattack that matches this cyber defence.

So here’s the research question for data curation: how can I tell, from the internal evidence, and partial external evidence, whether this particular document is legitimate (or is the legitimate version of a set of almost-replicates)?

6.5/7 US presidential elections predicted from language use

I couldn’t do a formal analysis of Trump/Clinton language because Trump didn’t put his speeches online — indeed many of them weren’t scripted. But, as I posted recently, his language was clearly closer to our model of how to win elections than Clinton’s was.

So since 1992, the language model has correctly predicted the outcome, except for 2000 when the model predicted a very slight advantage for Gore over Bush (which is sort of what happened).

People judge candidates on who they seem to be as a person, a large part of which is transmitted by the language they use. Negative and demeaning statements obviously affect this, but so does positivity and optimism.