Sunday, January 25, 2009

There is no penalty for being a predator.

Monday, January 5, 2009

need to work on that dream-recorder.

Last night was filled with bizarre and rememberworthy dreams. Walking around a park/castle/campus place and having lightning strike twice. Also the ridiculous amount of time spent walking around/interacting with people on/working at a campus earlier yesterday. So in-depth, these places are.

Monday, December 22, 2008

2 things, other things

Thing One:

"Crowdsourced Extensible Metatagging Mechanism" - essentially, content for your content, brought to you by the people. On your youtube video, you can have a whole ton of stuff - objects visible, words being said, people present, subjects discussed, references made, colours, links, images, audio clips, citations - literally anything.

Then make it searchable, and filterable.

Then implement a damn fine interface over it all.

Extensive notes in blue notebook that starts with times and schedules.


Thing Two:

Taxonomy for the Classification of Goals -- an afterthought to a Taxonomy for the Classification of Things I want to Learn.

Generally, there's enough things out there that I hear about / notice and end up thinking "I want to learn that" or "I want to do that" or "I want to be good at that" that I simply cannot remember them all. So it would make sense to formulate some sort of a taxonomy by which to classify them and then record them according to the taxonomy when I encounter them.

There is a possibility that I will start a new blog for this .. anyway, it's a high priority thing.


Other things:

Polypropagation was neat, so is the game (damn it I lost (this is not the first time I've been exposed to the game - that's just something I had to say - the game is hereused under an alternate definition)

That may be all for today.

Thursday, November 20, 2008

Duplicate File Checker

Multiple settings: Cautious, Sensible, Aggressive, Automatic, Custom.

Program which scans over a filesystem and creates a tree of checksums, and performs operations depending on the values.

Under 'cautious', lists all suspected duplicates (within 10% variance or so between sums) and allows the user to perform any actions like deletion, comparison, etc.

Under 'Sensible', prepares any exact duplicates to be deleted and lists any within a variance

Under Aggressive, deletes any exact duplicates and preps all within a variance to be deleted

Under Automatic, marks all within a certain range and then does a byte-for-byte comparison of each file with its suspected duplicate. Identical copies are deleted, near identical copies can be deleted or listed.

Under Custom, all options are user settable.

Could be run on two specific files, a folder, a number of CTRL+clicked files, or a whole filetree.

Let's see if it exists already

Managers are to Computers as Computers are to Computers

Long long ago, 'Computer' was a title given to a person who made calculations as a career. Often a second career - but this was something done as a chief timesink. Calculations were made of decimals of PI or calculations of logarithms or circumferences or areas or trajectories -- later on trajectories were much more common than the others. Slowly was it realized, and even more slowly was it implemented, that these tasks could be entirely automatic and performed by machines.

Now, to think of it - that's ridiculous. Why would a person ever spend decades of their life writing out numbers in a table? How mundane! How boring! How redundant! No human should waste their time on such a task. We have all come around to the mindset that computers in the machine-sense are superior. They do number crunching and a variety of other things - but they certainly beat out people at calculations.

So what I posit to you (finally), is not 'that there are other things we haven't heavily considered to be done by computers, which could be' - that much I hope is obvious. I posit that Management is a task best left to the machine mind.

Managers create timelines. Managers take in data and make estimations. Managers organize data and structures into reports and compact bits of information. Managers ensure that bills get paid, that people are compensated, that time is efficiently spent. Managers ensure that people play on an even field and all have work to do. Managers have very automatable jobs.

You can't be jealous of a Machine. It isn't trying to steal your job. It isn't gloating. It isn't going to hold a grudge against you, to fear being shown up, to fire you out of anger or spite or confusion. It isn't going to be lazy. It isn't going to sleep.

A machine manager would be efficient, timely, organized, even-handed, reasonable, never forget anything, and stay out of the way when it needs to. It can conduct communications with the team over IM or through emails. It could chair meetings - but organizing meeting agendas with small writeups might be more feasible and reasonable.

The machine manager need not be self-aware, nor an example of strong AI. It simply must perform its tasks as well as a human - that would take some study, some programming, and some standardization of I/O. An issue of hackability comes up, and a possible fix is to break management into several pieces which manage each other. This is difficult to enact with humans because greed, jealousy and paranoia seem to inevitably crop up in our actions - but a simulated 'group' of machine minds should be capable of monitoring each other easily without such risks. If one partition becomes compromised it is reset from a backup, or cut off and flagged for examination by a human. Something to that effect.

This is not just applicable to business; this is applicable to government. If our government is a tool of the people, why not make it a literal tool? Why have political parties and rhetoric and ideology? Why not simply have a system to hold referendums easily, and have machine managers present the facts in a stoic and pristine light so that we may absorb them, and provide informed decisions?

I think I would like such a society. I think such a society of machine-rule, under human control, has a great chance of being closer to utopia than anything we have yet done. -- And of course, it would have to be capable and expecting to yield to human control. As a tool of the people, as soon as the people vote to dismantle it, it must be prepared to do so in a manner unthreatening to our race.

Of course, I am describing ideals here, situations with no downsides - but that's the glory of a 20 year old's blog labelled 'ideas', hypotheticality is not only allowed and encouraged, but borders on required for inclusion here.

Anyway, goodnight.

Monday, November 17, 2008

Testing, testing.

Conceivable test strategy:

First, perform a fuzz test(/unit test? Something computationally intensive and time consuming) while performing a code review.

Next, check all direct inputs.
-Do one greater than max allowed,
-the max allowed, and
-one less than the max allowed,
-also flip of those for 'less than'/'min' -- this is all similar to
-Also conceivably maxint and minint equivalence partitioning
-input incorrect types
-null input, and
-maximum-length input.
-anything else suggested by the code review

Third, pairwise testing on all inputs.

Fourth, data comparisons

Tuesday, November 11, 2008

long term screen recording of professionals in the fields of software development, animation, modelling, architecture, art, etc.

Essentially, just record their screen and keystrokes over a long period of time to show the sort of methods they employ, to show their habits, to show the tricks of the trade that they may not even realize they know, or are special

These could be posted to various places as educational material - commentary in these could be cool too. Also could be used for employee review, for collaboration, a variety of things.

Have to be on non-NDA'd stuff, have to have a policy of "no judgement" or the person has the ability to edit pieces of their work out before it's reviewed.