Friday, November 27, 2015

A new truism! Truism 8

Picture of a map with two locations A and B and two routes joining them - a fast route and a slow one. In a narrative about going from A to B it is natural to say something like "I went from A to B but I went the slow way".
The use of "but" signals a truism to the effect that we assume actions taken are taken as efficiently as possible. How do I say this in notation? My best guess is that "good" has to be a sort of keyword; an adjective so general that, like verbs 'in' and 'at', it is allowed in describing a general truisms. So, taking us to the very edges of ethics and aesthetics, I propose:

Truism 8A: Actions are efficient:
(person->X) :: [(person->X)_/GOOD]

Truism 8B: Efficient actions have positive consequences
(person->X)_/GOOD :: [X_/GOOD]

Sunday, November 22, 2015

Does a visitor exist?

I don't believe Google Analytics when it tells me there are visitors to this blog. I never got a comment. Please leave one if you exist.

Wednesday, November 18, 2015

Implicit action versus implicit target for explicit action

The notation, luckily, is already able to distinguish what is needed. But it helps to clarify the convention (which I got straight during last night's insomnia);

X->Z is an explicit action where specificity is not needed for the action.
X->[Z] is an implicit action where no specificity is possible for the action
X-v->[Z] is an explicit action, specifically 'v', is taken with an implicit target.

"Jane throws the ball"...........Jane-throws->ball or Jane->ball
"Jane is ready"...... (Jane->[Z])*
"Jane throws"....... Jane -throw-> [Z]

So we can track the sub narratives on the way through:
"Jane is ready to throw the ball"
(Jane->[Z])* :: [Jane]-throw->[Z], ball

Friday, November 13, 2015

Deep AI on a chip

I do feel threatened by these announcements: "IBM Deep Blue...", "NVIDIA's deep neural nets on a circuit board", "Google Android announces open source AI".

Because I know those are stupid technologies that in no way represent "intelligence". They are brute force correlation engines which, without guiding "models", are stumbling around trying to find linear regressions in a universe of curved and non-convex categories.

What worries me is that with that much "snake oil" floating around, what would be the incentive to re-think things and get them right? On the one hand it is surprising what you can do with brute force. On the other hand I wonder if perhaps these announcements are more of a "Hail Mary" last-ditch effort to monetize a technology that the parent company knows is a waste of time. Maybe the companies are just dumping this stuff into open source with the thought that "maybe the public at large can find something useful to do with this dog of a technology....we'll focus our proprietary energies looking for something that does work".  - Fat chance.

Thursday, November 5, 2015

Speaking of cuttlefish carvings

Some I am more and some less proud of:
And...some lesser carvings, I used to test coloring