Editor's note: Andrew "bunnie" Huang is the author of Hacking the Xbox from No Starch Press. He works as a technologist for Xenatera LLC, and does freelance consulting work in many disciplines. He's currently involved with developing subwavelength photonic/electronic integrated circuits (with Luxtera) and silicon nanowire memory technology (with Caltech).
Andrew is a featured speaker at O'Reilly's upcoming Emerging Technology Conference. His talks are titled "Power to the People: Hardware Hacking for the Masses" and "Reverse Engineering: Basic Hardware and Software Techniques." In this interview, Andrew discusses flaws with the DMCA, the current states of reverse engineering and Moore's Law, what he's hacking now, and what his talks at ETech 2004 will cover.
Bruce Stewart: I'm curious what you think about the DMCA. Was it a major factor in the difficulties you had finding a publisher for Hacking the Xbox?
Andrew Huang: The DMCA was the major factor that made it difficult for me to find a publisher for Hacking the Xbox. The DMCA is a fairly expansive law with little solid precedent to define its scope or practical implications; hence, a cloud of FUD surrounds anything that could be dirtied by its waters. We've seen plenty of instances of corporations trying to push the limits of the DMCA to protect their interests (the Diebold voting machine case being the most recent one, in my recollection). Furthermore, the DMCA could be damaging to the current system of copyrights and patents in place to encourage and foster innovation. I could go on about litany of problems with the DMCA for pages, but basically, I can kind of see the need for a law like the DMCA, but I think the current implementation of the DMCA is not very good at all.
Stewart: Do you see any redeeming features to that law?
Huang: Well, the rise of "perfect digital copies" does present a number of business challenges to content providers. Creators of works should have some reasonable protections offered by law that enable them to make a living off of their efforts — otherwise, creators are less motivated to contribute their works, since the financial motives are supposedly diminished. Thus, the DMCA attempts to preserve the strength of copyright law in the face of easy, high-quality copying. However, I think the DMCA is flawed; it does not preserve the strength of existing copyrights effectively (music and software still get pirated en masse today), yet it enables corporations to eat away at certain traditional freedoms enjoyed by the public. If there is a redeeming feature of the DMCA, it is pretty weak.
Stewart: We have a long tradition of reverse-engineering in our technical society. Do you think the DMCA will successfully stifle that?
Huang: To some extent, the DMCA has already stifled reverse engineering. The fear and doubt that surrounds the activity has been enough to deter most technical people. The contrast between engineers in the U.S. and those overseas in countries without the DMCA, or DMCA-like laws, is quite stark. The technology marketplace is extremely competitive, and anything that slows down our ability to compete, such as depriving engineers of the unfettered freedom to reverse engineer competing products, may prove to have very undesirable long-term consequences.
Stewart: Besides hacking Xbox consoles, what else do you find interesting these days?
Huang: Oh man ... so many interesting things in the world! The more I learn, the less I know, and the more I want to learn. Right now, I'm playing around with silicon photonic circuits integrated with conventional electronic devices with a company called Luxtera. Bandwidth is a significant bottleneck in computer architectures, and the next big leap in bandwidth is likely to come in the form of some kind of integrated optical solution. It has been interesting applying myself to problems of this sort.
Another thing I've been playing with is silicon nanowires. I've been working with André DeHon (Caltech) and Charlie Leiber (Harvard) on some ultra-dense memory testbeds using silicon wires just a few tens of silicon atoms across.
I'm also actively playing around with various video-game-related hardware ideas. I can't talk much about it now, but if things go well, you'll probably hear about it through other channels.
There is also a lot of fun and interesting reverse engineering work still to be done. The rise of highly integrated embedded computers everywhere, combined with the introduction of the DMCA, has been prompting manufacturers of traditionally low-tech items to start using technological measures to limit the abilities of their products. These limitations are put in place for strategic business reasons that typically enrich corporate pockets at the expense of customer choice, pricing, or quality of features. This has, in turn, spurred a lot of interest from the public toward reverse engineering such schemes, since consumers desire cheap, featureful products without artificial crippling or barriers to usage.
Stewart: Can you give me some examples of the kind of products you're referring to?
Huang: Some examples of products today that have some kind of limitation being built into them that is regulated through an electronic technological measure include:
The mere existence of code-locked power supplies and ink cartridges boggles the mind. Vendors are very motivated to lock consumers into buying depletable supplies and accessories from the original vendor. After all, most of the profit is often made there. Yet consumers are always in search of a "better deal," which keeps the market competitive and lean.
The day may not be too far off when photocopiers come with mechanisms to detect infra-red or ultra-violet readable-only barcodes or RFIDs on plain paper to lock consumers into a particular kind of paper. Of course, such barcodes would be pitched as "feature enhancement and copyright enforcement mechanisms". For example, the copier can read the kind of coating used on the paper through the invisible barcode and fine-tune the toner application method to get better saturated dark regions.
So, there is always the good and the bad to the more intelligent consumeables and accessories. It's nice to have paper that knows how to program the copier for higher quality copies. At the same time, it would suck to have your privacy invaded or the copy-ability of a document lessened through some kind of poorly devised DRM or materials management scheme. In the end, one of the public's weapons for preserving their consumer freedoms and preserving the balance of technological power versus large corporations are the hackers at large who can understand and perhaps undo the potential abuses of otherwise well-intentioned ("good") technologies.
Stewart: One of your interests is in integrated circuit design. What developments do you find exciting in that area?
Huang: There is a lot of exciting stuff going on in that area all the time. There are a lot of things you've probably heard of — especially technologies related to scaling CMOS to finer geometries and lower voltages. Those are exciting, but I think even more exciting is the prospect of Moore's Law slowing down. The eventual leveling off of the aggressive exponential curve set by Moore's Law will introduce a huge paradigm shift in the electronics industry. A recent paper from Intel itself sets the end of Moore's Law around 2018, with maybe one or two more generations grudgingly following on beyond that date. That date is just 15 years from now; to put that in perspective, the x86 architecture is 25 years old, so this revolution is more than halfway over. Furthermore, the folks in my generation will be in their forties at the end of Moore's Law; arguably, it will be my generation that is going to be leading businesses during the post-Moore's-Law era of electronics. It is an interesting thought experiment to try and figure out what technologies will become important in that era and to prepare for its eventuality. Maybe it is too far in the future to tell, but some trends are available, if you care to look hard enough. I suspect the most successful of my contemporaries will be those who are prepared to meet the opportunities of the post-Moore's-Law world.
I also think that the developments in analog and RF circuitry are quite interesting to watch. This circuit design community has not been so blessed by Moore's Law, yet they have had to find ways to keep up with the increasing demand for bandwidth, precision, and accuracy. Communication bandwidth and latency are commonly acknowledged to be the Achilles' heel of modern processors, and the analog and RF circuit design communities are the front-line soldiers in the bandwidth battle. Coming into the fracas are the integrated photonics fellows. Their presence has not yet been felt by the electronics industry, but I personally feel that there is an undeniable tide pulling in favor of practical, integrated opto-electronic circuits.
Stewart: Tell us a little about silicon nanowire memory technology. Do you see this as the future for computer memory?
Huang: I think that if the basic technological challenges faced by silicon nanowire technology can be addressed, it could be a very potent and useful memory product, especially since it offers the promise of densities over 100 times today's densest DRAMs. Perhaps the most interesting aspect of silicon nanowire memory technology is the philosophy embodied in its assembly. It is one of the first technologies I am aware of that embraces stochastic assembly techniques at the architectural level. Silicon nanowires are too small to be reliably and economically manipulated or assembled into large arrays; thus, special coding schemes are used to enable deterministic memory addresses in the presence of high rates of wire duplication and/or deletion. The basic technology of taking an inherently unreliable process and "computing" around it in an efficient manner to create a reliable product is profound, and may come to play a larger role in conventional CMOS circuit-design techniques as linewidths scale and acceptable circuit yields become more difficult to achieve. In addition, the logical extension of being able to build dense memories with silicon nanowires is the ability to build dense programmable logic in the spirit of FPGAs. Thus, I may be unintentionally implying a limit to the abilities of silicon nanowires by referring to them in the context of my immediate memory-focused project.
Stewart: You'll be speaking on "Power to the People: Hardware Hacking for the Masses" at O'Reilly's upcoming Emerging Technology Conference. Can you give us a sneak preview of what you'll be discussing there?
Huang: This talk will be an extension to last year's tutorial on hardware hacking. The response I received from the attendees at last year's tutorial was truly overwhelming, and I had received numerous comments requesting a regular talk (with a larger potential attendance) on the topic.
My basic message is that hardware hacking is experiencing a renaissance: it's recently been enabled by various socio-economical factors as well as technological factors. Socio-economical factors include the slowing of the economy, which has allowed hardware hackers to "catch up" with the technology curve, and the increased demand for hardware hacking due to a rise in consumer-driven reverse engineering. Technological factors include the rise of better, cheaper designs and better diagnosis and fabrication tools, as well as the introduction of basic enabling technologies such as dense and affordable FPGAs (user-reconfigurable hardware).
I'll also be addressing some of the political and social issues associated with hardware hacking. There seems to be this mystique around the hardware hacker -- "Is that illegal?" and "Why would any legitimate consumer do that?" are questions I am asked quite often. I will probably spend a few minutes (but not much more than that) speaking to the bigger-picture issues.
I am personally very happy to see a revival in interest in hardware hacking. Looking back on the history of technology in America, I see a lot of hardware hackers, moonlighters, hobbyists, and enthusiasts playing key roles in developing the core technical competencies that have brought us to where we are today. I was concerned for a while to see ballooning admissions into CS departments at the expense of the EE departments in colleges. Without a fresh supply of EEs, where will the silicon infrastructure be to support all these CS folks?
Stewart: What talks are you looking forward to hearing at ETech 2004?
Huang: I think all of the talks sound really exciting. I'm betting that the discussion in the "Bad Device, No Biscuit" session will be lively, and not something I want to miss. I also think that the "Sampling the World" tutorial by Raffi Krikorian will be a must-see for hardware hackers; his tutorial will cover a lot of truly useful embedded-hardware hacking techniques, whereas my tutorial will focus more exclusively on specialized techniques for reverse engineering.
Stewart: Thanks for your time.
Huang: You're welcome.
Bruce Stewart is a freelance technology writer and editor.
Return to OpenP2P.com.
Copyright © 2009 O'Reilly Media, Inc.