» Conference Talks and Presentations

This talk covers the mathematical background behind elliptic curve cryptography, and the best methods for attacking ellipic curve cryptosystems. The elliptic curve variant of the discrete logarithm problem, and Pollard's Rho method for attacking the ECDLP will both be explained. Once the background is in place, the real-world method of implementing a parallelized method of Pollard's Rho attack will be described, and the feasiblity of such an attack will be discussed.

"Prowessful counterassertion, intercommon sunshining in synentognathous crossbreeds."

Rubi Con 5: March 29, 2003.
Slides in: postscript pdf

This talk will cover pseudorandom number generation, and how "randomness" can be defined. It will be oriented for those people who want a better understanding of what terms like "pseudorandom" and "bit entropy" mean, or are wondering why exactly it is that a computer can not generate a "truly random" number. A number of different aspects of pseudorandom number generation will also be covered, including ways of analyzing the strength and randomness of pseudorandom number generators, and explaining things such as what it means when nmap tells you that a PRNG is weak.

"/dev/erandom: The inner workings of a provably secure PRNG."

Toorcon 101: September 28, 2003.
Slides in: postscript pdf

Many efficient methods of generating "good" pseudorandom numbers exist in the literature of mathematics and computer science. One particular method of generating good randomness is to use "extractors": functions which will transform "bad" randomness (i.e. smaller ratio of entropy/data, or randomness distributed poorly) into "good" randomness (of a provable level of security) by an additional input of only a small number of truly random bits.

This talk will cover the mathematical background behind pseudorandom number generation, including concepts such as entropy and what "good" and "bad" randomness actually means. Once the appropriate background has been presented, the talk will move from the world of theory to that of practice, demonstrating how these concepts can be used for the purpose of pseudorandom number generation. Specifically, the current /dev/{,u}random PRNG for Linux will be discussed and compared to the new /dev/erandom PRNG written which uses these extractors.

"Computers without hardware; programming without coding."

In this talk, we will forget about all the modern advancements of programming languages dot Net and silicon, APIs and transistors, and look at the bigger picture. Rather than covering the work of the latest and greatest "computer security professionals" of our time, we will look at the work of people such as Alan Turing, Alonzo Church, Kurt Gödel, and Stephen Kleene. We will program without touching a computer, and play with languages that aren't used with compilers, linkers, or interpreters.

"Pseudorandom number generation, entropy harvesting, and provable security in Linux."

Black Hat Europe 2004: May 20, 2004.
C4 (Chaos Computer Club Cologne) Presentation: May 21, 2004.
Slides in: postscript pdf
Paper in: postscript pdf

Many efficient methods of generating "good'' random numbers exist in the literature of mathematics and theoretical computer science. One particular method of generating good randomness is to use "extractors'': graphs which will transform "bad'' randomness (i.e. smaller ratio of entropy/data, or randomness distributed poorly) to "good'' randomness (of a provable level of security) by an additional input of only a small number of truly random bits.

This talk will cover the mathematical background behind pseudorandom number generation, including concepts such as entropy and what "good'' and "bad'' randomness actually means. Once the appropriate background has been presented, the talk will move from the world of theory to that of practice, demonstrating how these concepts can be used for the purpose of pseudorandom number generation. Specifically, the current /dev/{,u}random PRNG for Linux will be discussed and compared to the new /dev/erandom PRNG written which uses these extractors.

Entropy harvesting will also be covered; the work on /dev/erandom prompted a number of improvements to the entropy harvesting methods used in the Linux kernel. The new framework for entropy harvesting will be demonstrated, and the advantages (specifically flexibility and extensibility) of the new method will be covered.

"Making use of the subliminal channel in DSA."

The Fifth Hope: July 10, 2004
Slides in: pdf
Audio in: mp3

This talk will focus on one reason why it's extremely important to verify the trustworthiness of your encryption programs. A number of papers about a subliminal channel in the Digital Signature Algorithm (DSA) used by the United States Digital Signature Standard were published more than ten years ago. This channel allows for undetectable communication via digital signatures. The subliminal channel is generally viewed as a method of legitimate but hidden communication, but it can also be used for leaking secret information (such as keys) in an undetectable way to anyone who knows what to look for. This presentation will show how this subliminal channel works and demonstrate - using a patched version of the GNU Privacy Guard - how to use it for both benign and malicious reasons: legitimate communication using the subliminal channel, and leaking secret keys with each signature.

"Subliminal channels in digital signatures -or- Why it's very important to verify trustworthiness of encryption programs."

DEF CON 12: July 31, 2004
Slides in: pdf

A number of papers about a subliminal channel in the Digital Signature Algorithm were published more than ten years ago, allowing for communication through digital signatures in an undetectable manner. The subliminal channel is generally viewed as a method of legitimate but hidden communication, but it can also be used for leaking secret information (such as keys) in a undetectable way to anyone who knows what to look for. I will present on how this subliminal channel works, and demonstrate using a patched version of the GNU Privacy Guard how to use it for both benign and malicious reasons, both of which have little to no prior implementation in encryption programs.

Learning OpenPGP By Example (Stupid Crypto Tricks with GnuPG?)

21C3: December 28, 2004
Slides in: pdf

The goal of this talk is to help demystify some of the internals of the OpenPGP standard, through example, so that others can learn from and hopefully continue the process. The current (free) open source implementations of the OpenPGP standard are easily better than many commercial solutions, as well as more readily supported. To do this, I will show off a number of the OpenPGP-based projects I've been working on lately, including: subliminally leaking keys in digital signatures; vanity key generation; extending the web of trust to ssh host keys; and maybe even some attacks against the keyserver network that I'll later regret showing off code for.

"applied cryptography? oh, I skimmed through that book once."

Shmoocon: February 5, 2005
capitalization verbatim from published abstract.
Slides in: pdf

this talk is inspired by the title quote, part of a response to the question "how much cryptography experience do you have?" normally, it wouldn't have been a big deal. in this case though, the person i was talking to was someone who'd just given a talk on his new web-based, new-and-improved system for cryptographically-secure email that is easy-enough-for-anyone-to-use. a system he'd written in his spare time and was plugging hard so that everyone in the world could feel safe that their email is secure.

riiiiiiight.

it's been getting too easy lately. want proof? i'm going to bring up a number of these systems that promise security, anonymity, authentication, non-repudiation, whatever other buzzwords in the general field of cryptology that happen to be big at the time. and then i'm going to show you how and why they're broken, along with the steps that could be taken to improve them.

i'll also show how systems that are very good can still have their weaknesses, which can range anywhere from mildly annoying to rather problematic. while i may regret it later on, i'll describe and demonstrate a few "attacks" which most people seem to have overlooked completely.

the end result? hopefully the audience will have a better understanding of the common mistakes that novice cryptographers make, and will avoid them in the future. hopefully people will have a better idea of how to determine what to avoid if they want to actually be secure. and hopefully i won't offend anyone too badly in the process.

"Breaking Down the Web of Trust"

22C3: December 29, 2005
Slides in: pdf
Video: torrent

Even with tutorials on the WoT and good trust policies the concept of "trust" can still be hard to grasp. Here we'll look at trust metrics, ways of using current trust systems better, and some non-crypto applications of trust.

The web of trust best known for its use in PGP is now used in a number of other applications and is established as a good method for doing non-centralized PKI. But how good is it? How does one define a metric for trusting a trust metric? We have keysigning parties and extensive tutorials on good trust policies, but a lot of people still don't understand the basic concept of "trust," especially when it is superimposed on the world of graph theory.

We'll take a look at the web of trust as it is currently used, including statistics on the PGP WoT and what that means in practical terms. And from there on, it's all about trust, including the trust metrics involved (and why they could be a lot better), and current "correct" practices for establishing trust (and why they could be a lot better). To wrap up, we'll look at the possibilities for doing other interesting (but non-cryptographic) applications involving trust.

"Building Communities in Self-Destructive Environments"

Northern Ohio Technological Advancement Conference 3: April 8, 2006
Slides in: ppt

Over the past number of years, I've put in a lot of time developing and maintaining resources that are open (in varying degrees) to the world. While I do consider these projects more than worthwhile, sometimes they can be hard to deal with. While there are always problems-- with time, financially, or with other resources-- it seems that the biggest headaches always come from within. In this talk, we'll look at a few communities of varying degrees of stability I've been involved with, the problems that have come up in and because of these communities, and the steps taken to help solve these problems. These will include things created both online (2600net, aculei.net) and in the real world (the Hacker Halfway House, the current yet-unnamed project I'm working on).

"Breaking Down the Web of Trust"

Hope Number 6: July 22, 2006
Slides in: ppt
Audio: mp3

The web of trust best known for its use in PGP is now used in a number of other applications and is established as a good method for doing non-centralized PKI. But how good is it? How does one define a metric for trusting a trust metric? We have key signing parties and extensive tutorials on good trust policies, but a lot of people still don't understand the basic concept of "trust," especially when it is superimposed on the world of graph theory.

Seth will take a look at the web of trust as it is currently used, including statistics on the PGP WoT and what that means in practical terms. And from there on, it's all about trust, including the trust metrics involved (and why they could be a lot better) and the current "correct" practices for establishing trust (and why they could be a lot better). To finish, Seth will talk about some of the many bad trust policies that have managed to become mainstream and commonly accepted, even by many self-described "computer security professionals."

"Your Name, Your Shoe Size, Your Identity? What Do We Trust In This Web?"

DEF CON 14: August 6, 2006
Slides in: ppt

The web of trust, as used in PGP, is a well-known system for establishing trust between people, even if the people have not previously met. Why does it work so well in crypto? The answer is simple: it's the same system that we all use on a daily basis when dealing with friends, family, relationships, andjust about everyone else we have to interact with. On the crypto side, however, there are a number of restrictions that limit the effectiveness of this trust network. While many "security professionals" say that they are mandatory, the system seems to work just as well without themŃ are they completely arbitrary? Here we'll look at a couple of these restrictions, focusing on the technical aspects of identity verification, and evaluate their effectiveness through a couple of real-world experiments.

"Online Communities and the Politics of DDoS"

Northern Ohio Technological Advancement Conference 4: April 28, 2007
Slides in: ppt

Like meetings in public spaces, online communities often are faced with the problem of policing themselves against people who only wish to disrupt things. This is even worse on the Internet, where bandwidth is cheap, botnets are easy to create, anonymity trumps fear of repercussions, and many people think denial of service attacks are completely legitimate ways to show off the size of one's e-penis. This talk will cover the problem of DDoSes to online communities: from why it's so common (technologically speaking), to the blame-the-victim mentality that perpetuates the problem (and how to fight it, both on a technical and social level).

"A Crash Course In The Math of Public Key Cryptography"

Chaos Communication Camp 2007: August 11, 2007
Notes in: pdf

This workshop is a lecture in the math background of public key encryption. (Bring something to take notes on!) All the prerequisite math will be covered, no prior background in mathematics is necessary. The goal of the lecture is to have everyone come out with enough understanding of some common algorithms, not just how but also why they work, that they may be able to reproduce and implement the algorithms properly.

The focus will be on the discrete logarithm problem, commonly used in the Diffie-Hellman key exchange, ElGamal encryption, and DSA digital signatures; factoring, which is commonly used by RSA encryption and digital signatures; and elliptic curves, used in variants of the DLP. The math covered will mostly include abstract algebra and number theory.

» Other Talks and Presentations

[ return to main page ]