- User:WillWare/Automation of science
- User:WillWare/Leopardboard DM368 hacking
I was the author of NanoCAD, an early predecessor to Nanorex's NanoEngineer-1 design program for nanotechnology. For a time I was a Nanorex employee. I have a website. Becoming a widower weakened my grip on reality, so for a while I regularly attended a Spiritualist church.
- 1 Evidence of an afterlife
- 2 Economics of cheaply copied products
- 3 Micropayments, reputation economies
- 4 Global warming
- 5 Electronics page
- 6 Building a hoverboard
- 7 Things I want to learn more about
- 8 Java (and other) things I want to learn about
- 9 Maybe I could make some instructional videos
Evidence of an afterlife
Spiritualism is a philosophy and religion whose core axiom is that those who've died continue to exist and remain sentient, and it is possible to communicate with them through a sort of interdimensional telepathy. Today it's a religion, but in the past it was simply an area of amateur investigation and research. My own experiences include being told things that people wouldn't know by ordinary means, and in a few cases, telling somebody else something that I shouldn't have known. Interesting stuff.
I have two problems with Spiritualism as it's practiced today.
- The vast majority of spiritualists have zero interest in making a credible logical case for their claims. That's a damn shame, because they probably could make one if they'd only try. I really wish I could teach them some statistics. The spiritualists of 100 years ago had significant debates with the scientists of their time.
- There is an idea called super-psi in the spiritualist research literature. If somebody tells me information that they can't know by ordinary means, maybe they read my mind, or did remote viewing. It's not necessarily the case that they communicated with somebody in the afterlife. This is one of those Occam's Razor things -- it's completely non-falsifiable, and it casts doubt on any pro-afterlife argument, but the a-priori probability for it seems kinda low.
The goal of Spiritualism today is to provide comfort to the grieving. It really is comforting to get a reading where something comes through. It's not solid proof of an afterlife, but you no longer feel like an idiot for entertaining the possibility of one.
So where's the good evidence?
Since my wife's death I have been reading about this stuff. Here are the impressive bodies of evidence that I've come across. These are the things with really solid data behind them.
- Ian Stevenson's work on reincarnation. Particularly the stuff with birthmarks that correlate precisely with wounds in the previous life. His work has been carried on by various colleagues, notably Jim Tucker.
- Melvin Morse's work on near-death experiences. He has managed to effectively counter-argue most conventional explanations in terms of anoxia, drug effects, and psychology.
The work by Stevenson, Tucker and Morse offers fairly credible hope for an afterlife.
Then there's the stuff that's also very interesting, but the results aren't widely accepted.
- Remote viewing
- The work on telekinesis at Princeton
- Gary Schwartz's work on validating psychic mediums
I was also astonished to learn that a guy I know has had a real near-death experience. He had an experience of leaving his body and getting information that shouldn't have been available to him.
Economics of cheaply copied products
Copyrighted materials like books and movies are instances of a more general thing. For a wide range of products, there is a very big investment required to make the first copy, but later copies are very cheap. The ratio is almost infinite for movies, which can cost many millions of dollars to produce, and can be shared for nearly zero on file-sharing networks. Another good example is pharmaceuticals, which go through a long and expensive process of development and clinical trials, but then the pills may be manufacturable for a nickel or a quarter.
It would be cool if each work of invention and creation could be generously compensated in the same one-time fashion in which it occurred, and the copies of the created work could then be available cheaply as commodity items. Ideally the inventor or creator would agree to receive a quick windfall, and in exchange, the invention or creation would immediately pass into the public domain. The copyright clause in the U.S. Constitution attempts something like this, because historically inventions were often lost because their inventors died before disclosing them.
Instead of a term of decades as we now have for patents and copyrights, the creator might receive all the projected revenue within a period of maybe a year or two, at which point the product or creation could become a commodity. This would be enormously beneficial to the consumer, as the prices of goods would fall more quickly.
Ransom model for big-budget movies
Let's suppose I'm the producer of the next blockbuster movie and it's 100 minutes long. I want to make one billion on the movie. I chop the movie into 100 one-minute segments and encrypt each one securely. (This can be done with 128-bit encryption keys, only sixteen bytes.) I post the encrypted slices on file-sharing networks like BitTorrent, and I don't bother to track them. They are unviewable until I release the encryption keys. Piracy ceases to be an issue.
I release a few encryption keys immediately, as the trailer for the movie. The corresponding slices of the movie are now available with all cinematic bells and whistles, at full resolution, Dolby sensurround, etc.
I release each encryption key under the ransom model for a total price of one billion. (Maybe the exciting parts are higher priced.) On average, I am looking for ten million dollars per key. The means for selling the keys might be an auction, or a private sale, or a more traditional ransom model of accepting donations. Customers need to understand that they are purchasing the public release of each encryption key.
As the ransom is paid on each encryption key, I publish it on my website and print it in a few major newspapers. I think this would be a viable model for a movie producer, and they would make a boatload of money with no energy wasted trying to fight piracy.
I don't see how a ransom model can apply to pharmaceuticals, since they cannot be meaningfully serialized. An entirely different model is required. A new drug might be purchased by the government and then released into the public domain. The rationale for this would be that the beneficiaries of the public domain release would be the American public ("the governed") and the government is simply acting as representative. Some portion of the federal budget would be earmarked for this purpose.
It would probably be smart to establish a prediction market to guesstimate a reasonable price for drug buy-outs. There is a lot of information involved and such a market could encourage the organization of that information. Creating the market first might be a good way to encourage the government to adopt that role.
Micropayments, reputation economies
In a system of electronic money, any piece of money is digital information, and copying that information cannot be prevented. No system of electronic money will be adopted if it fails to prevent double spending. The only way to prevent double spending is to involve a centralized server in all transactions. All participants must believe the central server is absolutely trustworthy, or they'll never use the system. So reputation is important.
If you're wealthy, setting up a bank is easy. The FDIC will be happy to certify your trustworthiness. Can a person of limited means become a micropayment banker without wealthy friends?
A micropayment bank that isn't recognized by some well-known authority might be able to earn trust using a reputation management system (like Cory Doctorow's Whuffie). Ideally such systems should be decentralized -- a single point of failure can fail for both technological reasons and due to treachery. There are a number of centralized reputation servers out there already including Slashdot karma and eBay ratings. Particularly with eBay, where the rating has tangible value, there are many schemes for distorting the rating.
A BitTorrent-style decentralized system would be very cool. It would allow people to establish micropayment banks. Perhaps it would some day ease the transition to Doctorow's picture of a post-scarcity economy, which seems like a very desirable long-term goal.
Google's PageRank system gives an example of a working reputation system. A link from one page to another is treated as a recommendation. The ranking contribution of a recommendation to the destination page is proportional to the ranking of the source page. It works out then that rankings are given as eigenvectors of a matrix whose entries are the recommendations, and there are parallelizable algorithms like power iteration that converge on rankings over multiple iterations. One presumes that Google runs a very large power iteration system to arrive at pageranks.
More thinking about reputation economy
Currently we use money based on the notion of property. Some of the assumptions are that
- many physical things can be exclusively owned by somebody (but not every physical thing, e.g. air or sunlight)
- a person's labor and thinking belong to that person until they give or sell it to somebody else, e.g. a work for hire
- some ideas can be exclusively owned - this idea is questioned by many
Briefly, I don't think any amount of robotics or nanotechnology or super-cheap physical goods will ever completely uproot the idea of ownership. Even a ballpoint pen or an apple is owned by somebody.
I expect that money will continue to exist in essentially its current form. There will probably be better and better social safety nets for the poorest people, at least here in the U.S., and there may be increasing use of "alternative" currencies like micropayments and local currencies, and I think both those things are laudable. But economics as we know it is not about to go out the window, regardless how much nanotech we bring on-line.
Credit scores work a bit like the notion of reputations. But a credit score is not transferrable: I can't give my friend a piece of my credit score, which might prevent him taking out a loan with a big institution, which would then lower his credit score. Meanwhile I might need to take out a loan from a big institution sooner myself, so in a limited sense my credit score is transferrable.
It seems to me that credit scores and money have different enough functions that you can't make them a single thing.
I saw An Inconvenient Truth last night. Gore presents the material compellingly and it's clear he's intimately familiar with the topic. There are a lot of ideas there, and they vary in objectivity. The most objective thing is that CO2 levels have obviously risen way beyond the natural cycles of the last 650,000 years, and over that same time, CO2 levels have correlated quite closely with global temperatures. A few of Gore's other ideas seemed more conjectural, but there is plenty there to worry about that appears indisputable.
- Greenhouse gas
- Climate change
- Climate surprise
- Retreat of glaciers since 1850
- Thermohaline circulation
- Shutdown of thermohaline circulation
- Kyoto Protocol
Mostly on this topic I feel despair. The production of greenhouse gases is driven by very large economic forces and wholeheartedly supported by the current administration. I don't think private citizens can make a significant dent, and big corporations will never be motivated to take any action.
One area where I might be able to contribute is in educational software, in particular, some kind of simulator of the various phenomena described in the movie. There was once a computer game called SimEarth where the player started with a prehistoric planet and tried to get intelligent life to evolve. I don't know whether Maxis worked too hard on the science.
I was an electrical engineer for many years, and a hobbyist before that, so I have an electronics page on Wikipedia. The gnuradio wiki is changing so I'm moving some of my information about radio theory to Wikipedia.
Building a hoverboard
I'm going to try to build a hoverboard. Here are some videos of people building and playing with hoverboards, and some other informational links:
- Long and detailed with info about design principles
- Here is a TV show in England where a guy built one and provided lots of instructions.
- The Human Hockey Puck
- Xinventions hovercraft
- Amateur Science website - the clearest instructions I've found about how to do this
These all run with gasoline-powered leafblowers. Some of them are designed so that the leafblower doesn't need to be destroyed. I have a leafblower, so here we go.
The clearest instructions I could find for doing this were here. I found some 1/4-inch plywood in the garage and cut out an oval-shaped board on the table saw. I cut out some thick plastic blue tarp for the skirt. All this stuff came from Home Depot.
Here's the leafblower, with its elliptical snout. I will need an elliptical hole in the board to put the snout in, so the underside of the board can fill with air.
Here are the tools for making the elliptical snout hole. It came out a little rough, but I hope to put something soft or foamy around it anyway, to help seal it, so I think I can tolerate this much slop. There is also a hole in the center, for a bolt that will hold the middle of the curtain to the top of the board.
So I got the thing put together. I connected the center of the skirt to the bottom of the board with a plywood disc strain relief. I stapled the skirt to the top of the board and duct-taped liberally all the way around the edge of the skirt. I put some duct tape around the elliptical snout hole to help the seal.
I don't think the duct tape around the snout hole did much good, the gap was still quite big. I may decide to sacrifice the leafblower after all, or maybe its normal snout is removable and replaceable with a hose whose other end could be liberally duct-taped to the board.
I got the thing working, after a fashion. I climbed aboard, trying standing, sitting, and kneeling positions. I had made the skirt too baggy, so the craft was much too tippy. Trying to just keep it balanced had me sweating like a horse. A couple of times, I was able to balance it well enough that it started to move frictionlessly for a second or two.
I need a tighter skirt, and a better seal for the leafblower. I might want to see about finding a lighter passenger.
I made the skirt tighter, and I improved the air seal between the leafblower and the skirt feed. For a couple of reasons, I can't successfully ride on the thing myself.
- The plywood base needs to be wider so I can balance more easily.
- The plywood base needs to be thicker and stiffer, so it doesn't flex under my weight.
- The air connections need to be better, the duct tape keeps leaking.
- It's OK to permanently strap the leafblower engine to the plywood base, leafblowers are cheap nowadays.
With the leafblower engine idling, it can hover its own weight quite nicely, and then I can kick it around a parking lot like a hockey puck sliding over ice. Here's a video of that happening.
Planning and prototyping are two contrasting engineering styles. A planner tries to think through every possible detail or failure mode first, plans out everything on paper, and only then starts building. A prototyper begins with a general idea of his goal, and builds a cheap quick version first. The balance between planning and prototyping is economical. The cost of prototyping is materials, labor, and the risk of problems that might have been avoided with planning. The cost of planning is time, and the risk of problems that can't be foreseen until you have a working model. Simulation is a half-way approach - a good simulator will alert you to problems you didn't foresee with pure analysis and design, but is cheaper than building a prototype.
Having built an initial prototype, I've learned some things.
- You need a larger, stiffer plywood base.
- The skirt does get scuffed up and requires occasional duct tape repair, and maybe some feet to protect it. It would be good if the skirt were easily replaceable.
- The leafblower should be secured to the plywood base.
- The air seals need to be good - too much reliance on huge wads of duct tape now.
Things I want to learn more about
Google Wave Gadget and Robot
- User:WillWare/Google Gadget Home Controller
There really needs to be a Google Wave simulator for testing robots without a constant cycle of redeployment. The app engine simulator already exists. Try coding this simulator in Google using the Wave API classes. Think about how to display a hypothetical wave in Python, and how to add new blips or edit old blips, probably in Tkinter.
- Guido van Rossum
Things I've learned about Gadget development
- Develop with GGE, deploy to your iGoogle home page.
- Browser caching: On each tweak, save to a new filename.
- foo001.xml, foo002.xml, foo003.xml, etc
- Use alert boxes and your browser's Error Console to watch what's going on.
- Also http://code.google.com/apis/wave/articles/gadgetdebugging.html
- MiniMessages might be good, I haven't tried them yet myself.
- Developing new Gadget code is laborious. Searching the existing examples for a good starting point that does approximately what you want is a lot quicker.
- Use the gadgets.* libraries
- makeRequest works only with publicly available servers on port 80
- If you're depending on a web service, requests won't work if the service hasn't yet been publicly deployed
- In my case, I'm co-developing a web service that will run on App Engine
- Follow design conventions -- applies to both Gadgets and Robots
Gadget Designer is a Windows program that designs DESKTOP Gadgets, which are (as far as I can tell) completely unrelated to Gadgets for web pages. The white rectangle is the "view" and you can change its properties including width and height.
Java (and other) things I want to learn about
- Spring Framework
- Dependency injection
- Inversion of control
- Martin Fowler article
- Test frameworks: Hudson, CruiseControl
- Heaps and stacks and other data structures
- IBM website, "Mastering Grails"
Maybe I could make some instructional videos
A good way to force yourself to learn something is to try to teach it to somebody else without embarrassment.
- Google Wave API
- GWT development (Windows, Linux)
- Gadget development (web)
- GPU and CUDA programming
- molecular modeling, mixed-mode modeling
- CamStudio is a Windows app (website)
- ScreenToaster is web-based
- Wink (website)
- Guncam is a $20 Windows program including a movie editor