Archive for July 16th, 2008

I propose one hard test for the progress of comp sci.  I’ve laid the ground work for a computational engine that can write late night talk show monologues as well as the human writers.

Do you think it’s possible?

Here’s my basic idea…code forth coming.


Some Basic Info

Mathematicas and Humor, a book by John Allen Paulos

Philosophy of Humor/Theories of Humor

Some useful mathematical theory


Joke Generator

Potential Ideas
Simple Program based on Replacement rules of Subjects, Relationships, Events

Simple Program of puns, word combinations, definition crossing

Simple programs and then an rich interface that uses and avatar or on screen talent to “tell” the selected jokes.  Would prefer it to all be computer based as we want to find out whether the “telling” of a joke contains a lot (most?) of the humor.
How to do this:

Prep: Create a database of common objects, slang terms, relationship descriptions

a) parse the news each night for subjects, relationships, objects, events

b) enumerate all jokes (basically sentence combinations) using replacement of subjects, objects, relationships with objects in the prep database.

c) run training algo against real monologues (what jokes are likely to be used based on past jokes)

d) tune it

e) create inflection and pausing algorithm that “tells the joke better”

We can exclude the use of existing monologues to train the algorithms and instead use an audience (internet visitors) to rate the jokes and monologues.  The algo can then learn what replacements, what structures, and what styles work best.  Though i think using existing monologues is realistic as most writers and comedians borrow from successful previous work to save a long, boring training period.

Exhaust all possibilities of jokes using replacement rules.  Then run this model against actual jokes used on late night television.

Analyze how many of the actual jokes we found.  Push this analysis to back in to give weighting to the generated jokes to predict late night monologues.

Can we ever replace monologue writers?

Read Full Post »

I’m interested in spelling out a general theory for gameshows.

Why: It’s fun.  It seems to be possible.  Its worth a lot of money.

What Makes You Think Its Possible: 50 years of gameshows and we seem to have a handful of models that work well.  Work well => stay on the air => audiences find it interesting => advertisers want to spend money.

What is the Structure of This General Theory:

  • Enumerate the basic types of games
  • Enumerate the gaming environment
  • Identify and Classify the game playing behaviors
  • Identify and Classify the game outputs
  • Enumerate the game playing strategies and classify their outcomes
  • Vet against upcoming shows
  • Summarize and conclude
  • Produce show based on theory

What Does My Intuition (previous experience) Tell Me Already:

  • There are 2 main types of games: skill and no skill (“luck”)
  • There are 5-8 variants on those two types in each One Player, Two Player, and Multiple Players
    • One Player
    • Two Player – sometimes the “house” is the other player
    • Multiple Players
    • Variant Breakout
      • Skill
        • Physical Skill – speed strength
        • Knowledge/Factual Skill – huge base of facts needed
        • Reasoning/Logic Skill – no prior knowlege required (Gambling strategy games included)
      • No Skill
        • Physics
        • Probability
    • We’re left with 15-24 actual game engines (there will be all sorts of environmental variations that turn these 15-24 games into a near infinite variety)
    • The outcomes/stakes at play can all be consider environmental factors (yes, they may change how the game is played but the rules of the game engine won’t change)
  • The Environment can be classified easily
    • Host
      • Famous
      • Non Famous
      • Involved in Game/Vs. Player
    • Audience
      • Big
      • Small
      • Live
      • Involved in Game Directly
    • Music
      • In studio
      • Added in after
      • Ambient
      • Part of Game
    • Lights
      • Involved in Game directly
      • Spotlight
    • Time Limits
      • Time limited game
      • Timing based game
    • Stakes/Risk
      • Risk to player
      • Risk to vs. player
      • Large Stakes (define)
  • Game theory will be useful in picking out strategies
  • List of Gameshows
  • The stimuli and operants and schedules can be identified and classified
    • audience feedback
    • lighting and music
    • real money present
    • touches from host
    • interaction with family
    • ….

What the heck would I do with all this?

Use it to come up with new shows and improve existing shows!

Get published!

While I work on this, check out all these papers.

Read Full Post »

The news is a maniacal scramble to make sense of the current financial situation around the world.

Predictions, ____ expert from _____ investment research firm, advice, soothsaying, modeling, bear vs. bull, Fed should do this, Fed shouldn’t do this…. and so on.

A truth I got comfortable with a long time ago but had reinforced over the last three weeks in my life.

Your model can only be as predictive as the thing you are modeling.

– Jason Cawley, Wolfram Research (and probably others…)

That’s a stupid statement, right? Duh.  I know that.

Really think about it though.  and then consider these things and your interpretation of them

  • weather forecasts
  • endless dow jones index reports
  • Political polls
  • compatibility tests in online dating
  • SAT scores
  • TSA profiling at airports
  • Annual budgeting for businesses
  • Or go through the latest in the news

All of these “indicators” attempt to predict complex systems/situations.  Those systems have to show some stability, some simplicity to ever give way to useful prediction (useful = do you get info you can use elsewhere and with enough time/energy to use it).

There is potential to get some local or short term prediction due to local or short term stability.  However, to effectively use that over time you need to be able to predict when that stability yields to a new pattern. That is where it gets difficult.

Yes, you can aggregate a lot of these indicators to produce some sort of statistical sample.  Usually though, you’re simply hiding the intesting stuff in the errors in your statistical model.  Washing out the outliers as “noise”.  The problem is, especially in the indicators above, it’s the outliers that matter! But I digress…

Point for financial pundits: national and global economics itself is complex. no simple model, simple statement, simple index can accurately model it. not even poorly.

Agree or Disagree?  Let’s have a discussion.

Read Full Post »

This is one fun little theorem.

Basically… if symbolic systems terminate (program halts/gives output), the terminating expression is independent of how the rules were applied.

You get “confluence” out of this.

You probably are thinking, “and so what does this have to do with my life?”

a) maybe nothing if arithmetic never enters your life (unlikely)

b) it’s extremely good to know when you use functional programming that you can get to the same answer with many different ways of writing something.  For good overview of functional programming, go here.

Read Full Post »