Japan Advanced Computing Initiatives 1989-1990

Looking at these old research papers and abstracts reminds me that a great deal of what accounts for news and major advancements is hardly new. These papers are 25 years old but focus on advanced computation, machine learning, autonomous vehicles, user interfaces, lots of sensors and highly distributed computing.Screenshot 2013-12-20 12.26.25

And even these research papers are not even that old. We’ve noted before that the basic windowing/mouse interface was invented back in the 1970’s and what inspired Steve Jobs in his UI design during a visit to the labs in 1979 or at least so the story goes.

Forty years later we are finally shifting gears into gesture, touch, voice and soon more natural and intuitive tracking and contextual commands.

There are some interesting terms and frameworks to be had from some of these old papers though. Ones I like in these are:

  • The separation of processing into “Logical” as defined by conscious, analytical, serial concentrated and digital and “Intuitive” as subconscious, synthetic, parallel distributed and analog. The terms might be a little dated and missing some aspects like probabalistic methods but it feels right to think of the process as having two sides.
  • Another useful concept is the use of “soft” processing when information is incomplete and very changeable versus “massively parallel distributed” for large and increasing amounts of data that is well connected.
  • The combination of these types of processing types leads to the powerful uses envisioned in terms of interfaces and intelligent machines.

The two papers in PDF form are available via these links:

Report of The Research Committee on The New Information Processing Technology

Japan Advanced Computing Interim Report


Semantic Networks for VLSI Design

This old IBM paper from 1989 was used during our discussions of how a semantic network database could be used for real-world applications.

The Abstract:

Tool developers for Computer Aided Design and Manufacturing (CAD/CAM) applications arc experimenting with some new techniques that could he useful to other kinds of applications. Hierarchical-design techniques and semantic-network databases appear to he a synergistic pair. Each helps the other to provide systems which have less complexity, more flexibility, and which avoid problem size limitations due to main memory size. In this paper, definitions are constructed for organizing the semantic network, and are used to express a general purpose hierarchical database schema complete with rules. That schema is also the basis for the inheritance of both rules and common facts. The rules may include application algorithms as well as data management functions. The tracing of a wire-net through a hierarchical design is Screenshot 2013-12-19 10.08.06demonstrated, focusing on the character of interface port instances and definitions. The semantic network is also used to track and identify versions of designs in the hierarchy. Then the synergism of versioncd hierarchical dnta manngement and a single data-model implementation of semantic networks is described. A list of aggravations which impede acceptance of hierarchical design also suggests resolutions to these problems.

Here is a link to the full report done by Robert Griffith at IBM.

The IBM IDEA Semantic Network Database from 1990

Before I left IBM in 1992 we worked hard to devise a brand new architecture and design for the flagship IBM mainframe. The semantic network database was one innovation that figured prominently into our plans.

Today we see big funding behind database technologies like MongoDB which makes me think back to these early days.

The paper and some example diagrams we used with executives is available here in PDF format: IDEAS: Semantic Network Database, IBM, 1990.

Screenshot 2013-12-18 21.54.32


Oh Hell! (The Card Game)

There may be a lot of reasons to say this but in this case it’s innocent – just a new card game I learned last weekend.

According to the mates in the hunting cabin this is a game played often in the Navy when they were sailing a few decades ago.

The game is best played with from 3 to 6 players.

It goes like this:

1. Players a dealt a number of cards each. The number is up to the dealer. Most often it’s between 5 and 11 but it can be 2 cards. The maximum is governed by the total number of players.

2. The dealer then puts the remaining cards aside and flips the top one up in the middle of the table. This suit is now trump.

3. At this point each player has to declare the number of “tricks” they will take given the hand they have, number of cards and the number of players. We’ll come back to bidding when we talk about scoring below.

4. Now the game begins with the player to the left of the dealer putting a card down. Each player in turn plays a card which must be either 1) the same suit or failing that 2) a trump card or failing that 3) any card.

5. After every player has put a card down the winner of that trick takes the cards and now leads with the next card. This goes on until all the cards have been played.

6. Everyone adds up the number of tricks they took and gets a score.


This is where the bid comes in. Basically you get 5 points for every trick you take if it matches your bid. So if you bid for 3 and end up with 3 you get 15 points.

However if you take more or less than your bid you have to subtract 5 points for each one. So if you bid 3 and took 2 then you get 10 points for the 2 tricks *minus* 5 points for being 1 off for a total of 5 points. If you bid 3 and took 5 then your score is 25-10 or 15 points. The scoring encourages players to bid somewhat aggressively and then use their cards and some strategy to get close to the bid.

Playing the game to 100 points yields about the right amount of playing time – at least when we did it three handed.

Strategy & Variations:

Anyone who has played hearts, spades or bridge will feel at home with this game. Being able to vary the number of cards dealt makes the game fun. The name comes from the inevitable “surprises” that come with the game.

One point to consider is that by dealing only 1/2 or 2/3rds of the deck means that players can’t rely on knowing which cards are in play which helps ensure that not everything goes as expected.

Having only played the game once so far it might be early to start mucking around with it but adding an element of passing cards before or after bidding might be fun. So would passing cards prior to the trump card being shown.

The dealer would be in control of this to keep the game moving. So they might deal seven cards and say pass two to the right either before or after the trump card is shown. Just adds to what is already a very fun game.