"History is just people doing things"
THE ABQ CORRESPONDENT
ISSN 1087-2302 Online Edition Number 263......October 2017
1985 for clients and contacts of
In the ‘80s we first worked at putting to use a neural net technology that allowed us to deal with uncertainties and probabilities not readily susceptible to handling by sensible if…then logic. Yeah, it’s kind of Bayesian. Our Adaptive Pattern Recognition Processing (APRP) was much scorned by real experts in the field; Microsoft, Apple, and the intelligence agencies told us not to bother them. APRP resided in about 4k of code embedded in an object-oriented operating system including the Savvy programming language running on an Apple II personal computer and on the IBM PC. It was practical and powerful, even running on the stone-age microcomputers of the era. A user won an Air Force contract that was anticipated to take 18 months, coordinating the functions of several incompatible systems. He says he did it in six weeks with Savvy. We played with the concept of The Memory Pit. Virtually all information storage systems such as libraries of books, databases, and encyclopedias, required the people creating them to make decisions about where to put what, classifying information so that people consulting them could home in easily on what they needed. The assumption was that the later searcher had the same view of classification the system designer had…which wasn’t necessarily so.* The Memory Pit allowed us to shovel information/data into it, just as it came, with no attempt to classify it in advance. The later searcher could enter a word or phrase, even a very long phrase, saying essentially “have you seen anything like this?” The Memory Pit would show what it thought was similar. Yes, you sort of do that in searching the web now, but Google, et al are looking for word matches. Pattern recognition is far more, looking for “strings of bits” representing text, images, sound…anything, finding what is “most like.” Our suspicion is that IBM’s Watson is The Memory Pit. It learns from experience what information has been associated even once with other information, forgetting nothing, improving performance steadily as it impassively soaks everything in without bias. A potentially worthwhile thing you could do with the The Memory Pit would be to pick certain events that have occurred…fires in industrial districts, for example. You could show The Memory Pit all available data from the period immediately preceding such events…weather, political news, phase of the moon, financial reports, news of epidemics, drug use statistics, proximity to perihelion and aphelion, the price of eggs in China…whatever there is, apparently relevant or not. If there is anything in common in the patterns of data just before warehouse fires. The Memory Pit will spot it. You might find it really helpful if The Memory Pit could say “Hey, I just noticed a pattern in the Goose Island area of Chicago that reminds me of warehouse fires in the past.” The warning comes not from analysis, but from observation. The Pit could be alert to such events as bridge collapses, landslides, outbreaks of violence, market crashes, wildlife die-offs, mad-cow disease outbreaks and the like. The computers we’re using these days are rather faster, with more capability than the Apple II. Oh, and this just turned up online…entirely relevant.
*A long-ago project to improve satellite navigation required quite specific knowledge of isolated islands that nobody had much paid attention to. One of those was Islote Pelado off the coast of Peru. Searching for anything conceivably helpful, Paul Honore’ and I wandered into the biology library at Stanford, which was undergoing reorganization at the time, with stuffed animals piled on tables and tucked into shelves randomly. Just poking around, we turned up a book called Birds of the Americas by a birdologist who had been on a guano-seeking ship that happened to visit Islote Pelado on Valentine’s Day in 1929. He noted that the island was surrounded by steep cliffs, making landing difficult (ah, the island didn’t much change size with the rising and falling tides) had little vegetation (ah, it stayed pretty much guano-color), didn’t contain enough guano to be worth harvesting and it was uninhabited (ah, its appearance from space wouldn’t change much with passing years). That’s what we needed to know. The point is that the author of the book thought it was about birds. The publisher thought it was about birds. the biology department at Stanford University thought it was about birds. We were probably the only people in the world who thought it was about islands. We found it by following a hunch. The Memory Pit is about hunches.
GLIDING THROUGH IT ALL
We speculated in Person Computing Magazine in about 1976 that it should be possible to lay out all we know of history in a big computer file, and then move through it all any way we choose. Call it Panhistory. It might be arranged in layers, with the current situation at the top…covering the whole world. The world wouldn’t necessarily be a sphere; its whole surface could be spread out, figuratively, as a plane. The distortions produced by optical projections of a spherical surface on a flat surface wouldn’t be a problem, because we could move from one latitude/longitude point to another, or stop being at one place, and start being at another, as we travel the WWW now. If the top layer of this representation of time and space is “this year,” the layer beneath it would be “last year. with images and reports of those earlier times…and so on, as far back as we want to go. There’s a good deal of argument these days about how long ago history began…and all viewpoints could be represented. Take your pick. Just as there are wonderful books with drawings of “a town” in different eras, showing the changing architecture, transport systems, clothing, vegetation, and activity through the ages, our panhistory could show all we think we know as we move freely through time and place, staring in fascination. This representation would be the subject of heated debate, of course, and people with strong views might wish to obliterate the representations with which they disagree…as in real life, if there is such a thing. Panhistory would be interesting, though, and might reveal trends and patterns we do not now suspect. At the time the notion first came up, the computers most of us had to work with were things like the Commodore Pet and the Radio Shack TRS-80. Even monster mainframes hadn’t the capabilities of the laptop on which this is being typed. The notion was fantasy.
Times have changed, Look at what a company called Descartes Labs is doing. This spinoff from Los Alamos National Labs intriguingly describes itself as “A data refinery to understand our planet.” Using satellite images of the earth in various forms, collected now for many years, they have concentrated on agriculture, making crop estimates, spotting blights, looking at the effects of fertilizers and pesticides, and providing never-before information about what’s growing. Notice one tantalizing illustration on their website that shows seven layers of observed data of different types. These fellows aren’t using DEC-8 computers, they’re imaginatively using systems with impressive muscle…but presumably not yet Quantum Computing that is just about upon us. Makes a body speculate again about Panhistory. Gee.
A media release we put out for UAVT Turbines, Inc, a while back was picked up as usual by a hundred or more “publications’ all over the world, most online. This includes websites for TV stations, all sorts of newsletters related to drones, as well as the regular aerospace media. The whole release is typically re-posted at these sites with no alteration of any kind, though some old-fashioned editors actually highlight specific information and create headlines directed to their particular readership. To our surprise, a publication called American Security Today, not only adjusted copy and headline, but found and ran an additional photo we had not supplied, featuring another company mentioned in the release. In this cut and paste era, that’s laudable, and when we dropped a bread and butter note to editor Tammy Waitt, she responded courteously, adding us to the mailing list for their daily newsletter. Not just clutter in the inbox, AST deals with more than pepper spray and bright flashlights for personal protection… getting into security implications of hurricanes, barriers at wide access control points, biometric identification, software vulnerabilities, perimeter lighting…even our engines for Group 3 unmanned aerial vehicles. Lots of fresh material relevant to our trying times.
This was passed along to us as “something to warm your heart.” It does.
The Correspo has commented more than once about using modern technology to increase the food supply. Here’s an item about greenhouse food production on a rather grand scale in the Netherlands. They’re not detectably doing “vertical farming;” this is planting directly on the polder under cover, and the results are impressive…not fancy salad greens, but crops like potatoes, getting twenty tons instead of nine tons per acre with dramatically reduced use of water.
ITEM FROM THE PAST
This item from 1995, twenty-two years ago, might have
been written just as appropriately today. (Note that there’s
no link here to the Forbes article mentioned in the first line.
In 1995, we were still printing and mailing the Correspo,
didn’t go online until the following year. Hard to imagine.)
Geoff Dolbear passed along a copy of a Forbes article reporting cheerfully on the stresses faced by the Reed-Elsevier publishing empire now that the Internet readily allows inexpensive distribution of technical papers and reports without the fuss and bother associated with publishing the work in ink on paper in academic journals. Anybody with something interesting to report can pump it out on the World Wide Web in preliminary form, inviting comment on the work. If the work stirs sufficient interest, the comments can be acknowledged and addressed in a reworked version of the document in later distribution, effectively providing “peer review.” Interested parties needn’t wait years to get news while the information is filtered through authorities who bless or condemn the work. (Somebody once observed that most people who receive Nobel prizes for their groundbreaking work were in danger of being tossed out of school by committees of experts who objected to that same disrespectful ground-breaking.) Steven Sester recalls that an official of a major scientific association with which he was connected demanded of the Wall Street Journal that all future articles dealing with technical matters be peer-reviewed in the public interest. Heck, in the 1970's one science-advisor-to-presidents recommended a “scientific supreme court” that would decide whether new projects deserved support and recognition or not, weeding out waste, and presumably blocking most new ideas from public view. The free-for-all Internet is slashing not only at the controlling superstitions of academic science, but at less formal publishing habits. One wonders if most technical trade publications are not similarly at risk. While this vigorous destruction of elitist social structures has its delightful aspects, it is not an unmixed treat. Somebody still has to pay for passing the information around. Advertising has supplied the fuel for free speech for many decades. Contrary to common belief, advertisers have little influence or even interest in the editorial content of the media they hope will carry their own messages to readers. They are interested in demographics and cost-per-thousand-exposures. How will advertisers use the Internet and its successors? That’s not yet wholly clear. Change is difficult, and there’s more change to come.
Well, things are still not wholly clear, and they are difficult.
The American Chemical Society was in September 2017
reportedly taking vigorous legal action against the operator
of a website that has been lifting technical papers from ACS
publications and others, and distributing them freely. One
may feel that it’s only right to punish such brazen
misappropriation, but the punishment requested could
restrict any reference to those materials by search engines,
for example. Putting expensive restrictions on searching
the internet seems rather like throwing out the baby with the
bathwater, but the ACS and others are much aggrieved,
urging strong measures.
making a case that academics should “take back control”
of research journals in the hope of reducing the cost of
disseminating research and of cleansing the field of
commercialism that may distort the science. Not everyone
is comfortable with having academics in “control” of
anything. Shenanigans in academia have been much
are fairly restrained examples. You can entertain yourself
for an afternoon by searching more colorful reports on
fraud in science. It really isn’t fair to pick on academia,
of course. All institutions, academic, corporate, political,
and religious, give weight to reports presented by their
members, quite unrelated to the intrinsic value of those
reports. Institutions tend also to protect their members.
It’s just the way things are, rationality aside; opinions,
interpretations, circumstances, and personalities lend
wholesome uncertainty to everything that is presented
as “knowledge.” No wicked intent is required. We can
be fairly sure that in another twenty-two years, these
And a tip of the hat to Joe Bosworth who died in his sleep August 10 at 75. Joe was a pioneer in the “personal” robot industry, founding RB Robot Corporation in the early 1980’s. Joe had read Robots On Your Doorstep (a book by me and Iben whose best review said it was “funky” and second-best said it was “bizarre”) and seeing something significant in it, he contacted me at Excalibur. This led to development of the remarkable Robot Control Language using Excalibur’s pattern-recognition-enabled Savvy programming language. Joe speculated once that the timing was right to stage a personal robot conference, and I replied with a nine-page memo discussing the possibilities. He then pushed through the first International Personal Robot Congress and Exposition…the IPRC, see this and this. It was all before its time. Joe became a good friend who showed up now and again, talking about interesting things. He introduced me to astronauts, pointed out surprising stuff in museums, and generally nudged robotics along. I was always glad to see him.
Copyright © 2017 ABQ Communications Corporation. All rights reserved.