SETI, Astronomy, and Cryptography papers
The paper Machine Intelligence, the Cost of Interstellar Travel and Fermi's Paradox,was published in the Quarterly Journal of the Royal Astronomical Society 35, no. 2, pp 157-175.  Arthur C. Clarke in 3001 called this paper "surely the most mind-stretching that the staid QJRAS has published in its entire career!"

I'm one of the editors for the book SETI 2020 and wrote several of the chapters.

I'm one of the unofficial champion of the opinion that the "Pioneer Anomaly" is not so anomalous, and can be well explained by conventional physics.  The main paper is "Conventional Forces can Explain the Anomalous Acceleration of Pioneer 10", Phys.Rev. D67 (2003) 084021.  See the paper from the on-line physics archive.   Also, I've put up a web page, with scanned references, for many of the thermal specific effects.

Here's a paper, now published in the electronic journal Contact in Context, on why aliens can not only detect our TV signals, but watch the shows if they want.  (Conventional wisdom is that they might detect the carrier of the TV signal, but they would not be able to decode the modulation.)  This paper argues that this is a pessimistic view of alien technology, and with likely improvements they can not only find the signals but demodulate them and view them.  So they can form their opinion of Earth from 'I Love Lucy', scary as that thought is.     Read the paper here

Here's another unpublished paper showing that if you are building a 'pincushion' style beacon using phased array technology, then amplitude control is not so important.  Fixed amplitude loses only about 1.05 db.  Read paper here  

Here's a paper, now published by Radio Science, that describes A Scheme for a High Power, Low Cost, Microwave Transmitter for Deep Space Applications.  This contains, as an appendix, the argument of the previous paper. 

Here's a paper that describes a way we might go about transmitting for SETI, and shows it is relatively cheap. Entitled A Low Cost Self Consistent SETI Strategy, or Talking and Listening is Not Much More Expensive than Listening Alone, it also points out that we may well get signals first from those most motivated to send, not those with the most unbiased or generally held viewpoints, in analogy to missionaries on Earth.

I enjoy cryptography problems.  There's a tongue-in-cheek language, called Malbolge, designed to be difficult (or perhaps impossible - it's not yet been proved to be Turing complete) to use.  By treating it as a cryptography problem, I've made some progress.  See this unpublished (with excellent reason) research here.

While playing CandyLand with my 3 year old son, I started wondering how long the game can be expected to last, what's the advantage of going first, and so on.   So I did a Mathematical Analysis of CandyLand (Classic version).

I think Wikipedia is an excellent idea, and have contributed a few pages as user LouScheffer.  Examples are  parasitic extraction as a professional contribution, and pages on sample rate conversion, and fictional papers from the Sherlock Holmes books as hobby level contributions.


The Buffalo Chip

In 1975-1977, I worked at Hewlett-Packard in Loveland, Colorado. (This was before HP split into HP and Agilent).  The project I worked on was the HP 3582a, an audio frequency digital spectrum analyzer.  This project had the code name "Buffalo", which stood for "Barely Usable Fast Fourier Analyzer Like Object".  (Management decreed projects must have names, but not how the names were selected...) Naturally the chip Lynn Schmidt and I designed for this project became the "Buffalo chip".  For those who do not know, the normal meaning of "buffalo chip" is the dung deposited by the buffalo from the large amount of grass that it eats.  You can see a photo you can examine from all sides here, courtesy of the National Park Service.   You can even buy one here!  

The layout department folks working on the project (Kathy Mazzotta, Yvette Norman, and Sandy Shertzer) drew up a picture of a buffalo and included it in a blank portion of the chip.  Every time the chip was revised, one more "chip" was added behind the buffalo. You would think this is about as obscure as it is possible to get - a microscopic image, sealed inside a low volume chip, used only in a specialized scientific instrument.  However, not only did someone find it, they took pictures of it and published it on the web!  See the pictures here from the Silicon Zoo at FSU.  It also has pictures of the heart that was on all the chips made at the Loveland, Colorado HP plant.

At least once the buffalo actually served a useful purpose.  When the first revision of the chip came back, we could not even power it up since it had a power-ground short in one of the output drivers.  To be able to power up the chip, and do at least some debugging, we decided to try to cut the power line to the bad bank of output drivers. We attempted this using a laser zapping system normally meant for adjusting laser-trimmed precision resistors.  However, since the laser was normally used to zap relatively thick and refractory nichrome resistors, we needed to figure out how to set the intensity to cut aluminum without digging a crater in the chip.  To do this, we used the buffalo as a target, adjusting the laser power until it created holes a little bigger than  the size of the line we wanted to cut.  So we might be the only folks who ever went buffalo hunting with a laser.

The Doug Tygar Practical Joke

Doug Tygar (currently a professor at Berkeley) is very smart, but at least when he was younger he was unsure of himself.  So when he finished grad school, and wanted to become a professor, he applied to Berkeley, MIT, Stanford, University of Illinois, CMU, Maryland, Texas, UNC, CalTech, NYU, Toronto, USC, Chicago, Princeton, UCLA, Washington, Columbia, Purdue, UCLA (different department), Wisconsin, Davis, Rice, UC Santa Barbara, Yale, Georgia Tech, Rutgers, UC Santa Cruz, Indiana, Stanford (different department), and UC San Diego.   He asked me to write 31 recommendation letters, one to each of these schools, and fill out each school's recommendation forms (probably an average of 5 pages each).  All this work was completely unnecessary; Doug already had a great track record by the end of grad school, and anyone who Doug knew that at most a few schools would suffice.  So we (myself, Ron Ellickson, Rodger Hughes, Rodger Scott, Leslie Fong, and a few others) did this:
About 2 days later, I got the following notes on my desk (this is before voicemail...)  10:00 - Doug called.  10:10 - Doug called.   10:20 - Doug called.  10:30 - Never mind.  By this time Doug had called the schools, who assured him there was nothing particularly outrageous about his reference letters.

Doug, of course, was accepted as a professor by the top choice on his list.
Erdös Number
My Erdös number is at most 3:
I asked Jerry Grossman,  who keeps the web site on Erdös numbers, about whether patents should be included.  We both thought they should, since they are collaborative work, they are published, and they are peer reviewed.).  If patents are not considered, then my number is at most 5, though the chain - myself, Carl Sechen, Alberto Sangiovanni-Vincintelli, Robert Brayton, Alan Hoffman, Erdös)
Rejected papers

Here's a paper that was rejected (by ICCAD 2003), but I like.  The basic idea is to perform error correcting codes on entire functional units instead of bits.  It's not my area of specialization, but I could not find the idea in the literature anywhere.  The closest match is work on "Algorithm Based Fault Tolerance", but these  papers usually worry about making specific algorithms fault tolerant (extra redundant row and columns on matrix ops, for example).  This paper's idea will work on any operations that are composable, provided you are doing lots of them.
Detecting and Correcting Errors in Functional Units Performing Composable Operations

Work Papers
Plus, I have lots of work related papers, but most of these are available in the IEEE or ACM on-line libraries.  Here's an example:

Explicit Computation of Performance as a Function of Process Variation, from TAU 2002

However, papers from TAU 2004 on are not published, since this is a workshop and not a conference.  The PowerPoint presentation from TAU 2004 is available from the TAU web site. but the corresponding paper is not.  It is available here, as The Count of Monte Carlo.  Note that if you are looking for "The Count of Monte Cristo", the famous novel by Alexandre Dumas, then you have come to the wrong place. 

Likewise the paper from TAU 2006,  "Why are Timing Estimates so Uncertain? What could we do about this?" is available here.

Also there are some really, really ancient papers, from before the web.  Here are scans of some of them:

LSI Design Verification Using Topology Extraction, Louis Scheffer and Ravi Apte, 12th Asilomar Conference on Circuits, Systems, and Computers, Nov. 1978, pp. 149-153.

A Methodology for Improved Verification Of VLSI Designs without Loss Of Area, Louis K. Scheffer, Second Caltech Conference on VLSI, 1981, pp. 299-309.

Design and Simulation of VLSI Circuits, Louis K. Scheffer, Richard I. Dowell, and Ravi M. Apte, Hewlett Packard Journal, Volume 32, Number 6, June 1981, pp. 12-16.

A Constrained Design Methodology for VLSI, Lou Scheffer and Mike Tucker, VLSI Design, May/June 1982, pp. 60-65.

Hierarchical Analysis of IC Artwork with User-Defined Rules, Louis K. Scheffer and Ronny Soetarman, IEEE Design and Test, February 1986, pp. 66-74.  This one is based on the paper from DAC 84 or 85.


Return to our home page