Create a radio that receives all frequencies of interest, convert it to digital, filter out the music and junk, and convert the remaining speech to text.

Sculpture instances.

- Vision
- Joy
- Despair
- Hope
- Faith
- Struggle

Studies conducted by linguists on the way people speak - syntax, vocabulary, etc. as a reflection of how they think. Need to find more information on these topics more readily. Just like everything else, analyzing it can be distilled to a simple process, and need to figure out how.

The instructor is not a person. People have varying teaching methods, resulting in students with varying levels of understanding depending on who taught them. The teacher will take on a secondary role as clarification instead, by answering student questions.

Students will learn from video courses. Each student will have a touch screen desk and learn from an animated video lesson. This allows student control of pace. With videos, it eliminates the potential disadvantage of bad teachers. In addition, videos can be split tested for teaching effectiveness. By modifying certain parts of the instructional videos into A and B categories, and analyzing the scores of the students who took the A and B tests, we can compare scores and see whether one particular teaching method was more effective. Obviously there are factors that will affect this, but this is a generalization. With A and B split testing, the lessons can be continuously improved. By the end of the semester, if a either group A or group B shows a statistically significant score from the other, their score will be modified.

The lessons should also be based on a question and answer format. All lessons currently consist of a mess of facts presented with no explanation of their applications, significance or anything. They are presented as nothing more than facts to remember, and this makes it difficult.

Selectively breed beans for maximum size until x(?) generations later, they are tomato sized super beans so large that a couple could be a meal. Would be very cool. No idea how many generations it would take. But once I get a pad I'll probably start testing until the end of my life. Hopefully the beans will at least have doubled in size by then.

Why beans? Beans are healthy and nutrient dense. They're delicious. But they're small, so they're perfect for breeding.

Because the first strike can easily mean a win for either side, determining the best ways to allocate the resources is important. The top players are skilled simply because they have fine tuned it to a point.

In all RTSs, you can basically harvest resources or build attacking units. Because the rate of resource acquisition and creation of new units is essentially constant, simple flowcharts can be created to determine the various combinations of units for optimal expansion or offensive capability.

Just another idea I will pursue when I've got loads of cash, time, an F430, and a nice McMansion.

Semi related article I read a while back. Neither of these methods of creating an AI seem to be reasonable. Both appear to require an incredible amount of input that would be unreasonable, and if they were created, they would not be capable of reasoning in a timely manner.

Programming basic common sense facts would take ages. It's not reasonable either. My idea would be to create an AI, and because of the lack of physical limitations, the machine can be trained under controlled conditions and learn all of these faster than it would take to program the facts in because it has the capability of learning.

The other idea seems even more complex than Mindpixel.

My line of reasoning assumes the determinism is true. All events are caused by preceding events that prime the action to occur, no matter how "in control" of the event one may seem to be. Therefore, there is no free will. Free will does not make sense. All decisions are reached based on certain considerations, as they are not random. What are the considerations? Obviously the inputs. But the resulting action is an output determined by inputs, therefore, intelligence is nothing more than an extremely sophisticated program which makes decisions based on inputs. Thus AIs can be created once all the decision making variables are in place. A newborn is capable of developing into an intelligent individual capable of reasoning and learning, which is what I assume the ultimate goal is - a machine that can reason and learn like a person. So why not simply program a machine with the same capabilities of a newborn and simply train it. Learning is determinism in a sense.

But there's also the obvious problem of our wonderful new robotic friends deciding to kill us all for whatever reason. The 3 laws cannot be hardwired. If it is a true AI, it could potentially override the "laws" because it has the capability of reasoning. Humans have things that are hardwired, such as habits and such determined by genes. However, these are nothing more than a "default" setting, capable of being overridden by the mind with sufficient effort or manipulation. Therefore with this hypothetical AI, if the human mind is fully modeled there is a significant risk that any 3 laws will be overridden if certain undesirable human characteristics are included. Therefore specialized "dumbed down" robots would be required for all tasks, otherwise there would be significant risk. Maybe only several AIs can be made that are not stripped of their capability to learn, and even these must be fully contained.

The most basic program of intelligence is simply "randomness". Random action, thought generates input into the environment, and possibly a response, allowing learning to occur.

Most Difficult Areas To Address
- Pattern recognition required for learning. Visual, auditory, physical, taste, balance, etc.
- Factoring the endless internal variables that affect a situation
- Determining which variables are even necessary for a task, such as emotion.
- Determining how the variables are connected
- Determining how closely the variables should mimic the human mind
- Creating the variety of machines with and without particular features

-Random Motion
|-Reflex
|-Learned Program
|-Preprogrammed

Often I don't plan things well. I have a terrible habit of jumping into things, forgetting the planning stage altogether time and time again. It's an irritating habit, and often I'll end up wasting time and such. It is also possible to refine every task down into a simple methodology for easy execution, even brainstorming. Hit upon a decent way last night for new webapp ideas.

Using the main trends of the internet, combining them in different ways and applying them to any niche yielding easily generated unique ideas. Came up with two actually good ones, but I was lazy and didn't get out of bed to write it down and forgot.

- From centralization to the collective, harnessing the power of masses through the internet
- From centralization to the individual
- Shift from local software to webapps
- Freemium, new pricing models
- Online communities and interactivity. Nothing needs to be done alone anymore in a virtual sense.

Also, Grand Map. A massive chart that will organize everything in life into a giant map flowchart of some sort.

This idea kind of falls in line with my other ones, which include optimization and data correlation - the Demographic Correlation Web. This hypothetical tool would link all the possible variables regarding people together, for example, weight, hair color, age, gender, location, siblings, interests, dislikes, etc. This tool would probably primarily be used by advertisers.

As far as I know, advertisers can only make educated assumptions and guesses at the interests of particular demographics and whether the demographic would be interested in a product, service, etc.

I'll just use the example of Kraft Cheese Nips, because I have a box of this nasty stuff on my desk. (Cheese Itz are better.) So say they want to advertise Cheese Nips on television. They have no way of knowing what kind of people like Cheese Nips. Other than making educated guesses like "kids ages 10-20" or something like that, and slapping their Cheese Nips ad on Cartoon Network or something, they have no way of knowing whether this is the only way they can successfuly target their ad. Perhaps there is a geographic targeting opportunity they are missing. It's said that Hawaiians like Spam. Obviously this can be attributed to historical reasons probably, but the point is that the tool would open up new methods of targeting. It could potentially eliminate the trial and error that goes into advertising, because there would be a database that relates different factors together.

Interests
Music
Movies
Likes
Dislikes

Physical
Age
Weight
Gender
Hair Color
Siblings

Location
Geography

The extent to which I'm taking the idea is obviously possibly unreasonable, but the data would be anonymous, so it could work. But it would be impossible, since the data would be impossible to collect. But alternatively, perhaps I could custom code a tool (with coding skills I do not have) to do exactly this (an outrageous plan that would ultimately do nothing?), using resources I have like Facebook and MySpace, both of which have really cool demographic targeting features. The tool would run through all of the possible combinations, and in theory it would spit out pure gold and allow the person who created it to retire to live the good life.

This could be carried out. Probably would need a very powerful, dedicated computer, some h4x0r coders, and intestinal fortitude.

But if there is a tool like this, I'd like to know about it...

A while back I wrote about a crazy scheme to generate perfect music. Thinking about it for a while, but never really got anywhere with it. Time to start the "research" phease, if you could call it that. I probably won't make it past here since I don't have a significant background in music theory, so I won't be able to analyze the more important "musical" aspects of these songs.

So here's a summary of the idea. Music is just a series of different tones. When played in succession and continuously, you've got yourself a trippy techno beat, sappy love song or one of those songs Bill O'Reilly regularly complains about. So the idea, inspired by the "perfect face" algorithm, is that universally (95%-99%) pleasing music must have some sort of similarity that makes them pleasing to such a large majority of the population.

For now the experiment will be limited to just one culture - American. Obviously we can't throw in some chart topping Bollywood beats or something. So the description "universally appealing" will indicate a song that is a chart topper on the Billboard Hot 100. Here's some more on that. The chart topping songs are selected by sales and airtime. Initially I thought this would be a problem, as only a particular type of person would call up a radio or buy a CD, but it's probably not a significant demographic group that it would interfere with the results.

So here I've selected some songs that have reached the #1 spot on the Billboard Hot 100 to compare.

Flo Rida
Right Round

Crank That (Soulja Boy)
Soulja Boy Tell 'Em

Whatever You Like
T.I.

Viva La Vida
Coldplay

Just Dance
Lady GaGa Ft. Colby O'Donis

Low
Flo Rida

There's a lot of hip-hop on the Hot 100, but I drew extra songs from other genres. Here's the game plan, or the aspects of the songs to be compared.

Song Name
Number of weeks on chart
Genre
Theme (Ex. What is the song mostly about?)
Synopsis (Ex. Full blown summary of the lyrics)
Tempo
My opinion of the song (Ex. Obviously the most important variable here. Might as well disregard everything else...)
Syncopation
Chorus (Ex. Yes or no?)
Time Signature
Structure (Ex. ABCDABAB)
Artist 1
- Gender
- Voice Range (Ex. high, medium, low, for lack of better description).
- Digitally augmented?
Artist 2
- Gender
- Voice Range (Ex. high, medium, low, for lack of better description).
- Digitally augmented?
Artist 3 etc
- Gender
- Voice Range (Ex. high, medium, low, for lack of better description).
- Digitally augmented?
Instrument 1
- Digitally augmented?
Instrument 2
- Digitally augmented?
Instrument 3 etc
- Digitally augmented?

I had more features to analyze before. Can't think of any more at the moment.

I'm currently taking AP Psychology. I posted a random conjecture (amusing that I categorized it under "science" as well...) a while back. Although I've only read a chapter in the book, I've gotten a bit more perspective on psychology and consciousness. Surprisingly my idea happens to be pretty much the same as Skinner's idea - that people are not actually in control of their actions - there is no free will. All actions are determined by a combination of nature and nurture.

Here's a few links for some background info on this -
http://en.wikipedia.org/wiki/B._F._Skinner
http://en.wikipedia.org/wiki/Free_will
http://en.wikipedia.org/wiki/Radical_Behaviorism
http://mb-soft.com/believe/text/determin.htm
http://en.wikipedia.org/wiki/Artificial_intelligence

Some of these topics merge with philosophy however, and start getting very confusing, as does the term "consciousness" itself.

But the most important part I forgot to mention about the idea is that because conscious functioning and learning of animals is just based on simple input output like machines, which is what determinism (or at least my previous idea) is, artificial intelligence could easily programmed with any type of code, since all code supports if then statements. All you would need to program any animal is to know every single variable involved, and then provide every variable a certain significance value to determine how important the variable is to the overall equation. For example, if you gave the "hunger" variable a 1/10, the algorithm would not consider than an important factor in making decisions. These default values would be your genes. However, because biological determinism simply cannot be possible, the default values can be overpowered by "nurture". At this point it's too complicated without some crazy diagram or something.

Once you can successfully mimic all the variables on a human level in a machine, all the work is pretty much done. Animals are kind of like OSs, but they are programmed to learn. So once you've got the "OS" ready, it'll just start programming itself as long as you teach it.

Very disorganized idea overall, because there were too many things I wanted to cover. But at least I will remember it.

This is just an idea for now which I will come back to once I've mastered a few other things.

Search engines are outdated (yeah... lame intro. I've never been great at writing...). Every other web based service as improved and developed over time. Web apps have become more like desktop apps with fancy animations, drag and drop and all sorts of great features. But search engines are still all have the same features they had when they first appeared.

All search engines are pretty much the same in terms of how the results are delivered. They're all static pages with a list of text results, search box on top, and additional pages on the bottom. Obviously there's a great deal that can be improved. Some search engines are much more innovative in the way information is presented, such as ask.com and the much hyped cuil.com.

ask.com has many notable features. It's the only search engine (that I know of) that actually detects your screen resolution and takes advantage of wasted space. All other search engines simply leaves swaths of white if you are one of the many who have higher resolution monitors. Ask displays helpful suggestions to narrow and expand your search on the left, and, if you've got enough pixels, widget like blocks with news, images, dictionary, synonyms, and encyclopedia terms related to a search term. In addition, instead of scrolling the engine page, search box and all, the navigation acts like a toolbar, allowing easy access. The only thing that could be improved is if they extended this concept to the search pages. It also has a very useful preview feature that allows you to scope out a page before clicking. It's also got a useful search history, which is useful, for lack of better ways to expand upon the idea.

cuil.com is that much hyped new browser created by ex Googlers. The search is pretty bad, but I'm only focusing on user interface for now. Cuil has a refreshing user interface and is much more fun to use than boring text results like the other search engines. Its design seems rather forward thinking in that it takes advantage of horizontal screen real estate more than Google, Yahoo and Live. After all, if you can set options to show results as text or icons on your computer, why not on the web? It functions much more like a local disk search in the way the results can be displayed, as results can be sorted into columns. However, it does not allow the option to change into text results, which can be a problem.

There's a few other sites such as Yahoo Alpha and Searchmash that implement a customizable sidebar, but the customization options could be much improved.

But really what these search engines have done is just the tip of the iceberg. So much more can be done to improve search. So here's a list of crazy ideas of what a truly modern search engine should have. I would draw some mockups and post them, but I don't have a camera with me.

1. Stationary Navigation and Search Box - When you search on the computer and scroll through the results, does the navigation disappear? It stays on top so you can easily modify your search. Cuil.com gets this right.
2. Modifiable Sidebar - Ask.com gets this right. Live.com tries, but the implementation isn't as good as Ask's. But a sidebar could do so so much more. Essentially you could add custom search widgets. For example, if you want a glance of images along with your results, add an image widget. Users can also submit custom widgets for their sites or for well known sites like Wikipedia. Some other widgets could be definition, antonym and synonym, encyclopedia, or news.
3. Easily Accessible Advanced Search - All search engines create a huge disconnect when users want to select advanced search. Most lead to another page, ask.com creates an irritating popup which disappears. The problem is that for all of these advanced search features the result is that on the new page, the advanced search box just disappears, leaving less skilled users with custom tags like -,"", +, site:, link: etc. Plus few are familiar with all of the modifiers to fully take advantage of the feature. Advanced search should slide down from under the search box, much like Vista's advanced search, and it should remain until the user chooses to close it.
4. Search Suggestions - Every search engine needs one.
5. Search Other Sites - Suppose you search many different sites, you can add options to search other sites from within the search engine.
6. Results View Option - Ability to switch between tiles (screenshots) and list view. I'm tired of text.
7. Some Sort of Self Improving Cloud Thing - Like Apple's Genius feature, but with search. Basically instead of improving a music selection algorithm (lame), it would improve search. I'm not quite sure how this could work, but it's possible.
8. Stationary Advertisements on the Left - It would improve visibility and clicks, so everyone benefits. Results would stay in the middle.
9. Change Results Per Page - Use a slider like the magnification on in Office 2007.
10. Notepad - Live, Google, and Tafiti have one, but each only works for a particular search, such as images or web. This feature should be available for all kinds of search, and could be stored along with a web history.

Just some ideas. Please tell me your thoughts!

But now that I'm finishing this post... if I had the skill I would totally start my own search engine with these features. These are (I hope?) great ideas that would really improve searching. Maybe it could become the next Google lol.

We are not actually conscious in the sense that we actually know it. Organisms are analogous to machines. In fact consciousness is simply the rapid execution of endless yes or no decisions, the same method a computer runs. I will use programs interchangeably with memory to liken organisms to computers. While programs are not exactly the same as memories, memories as used as plans to act, which is why they are similar.

When we are born, we have a body, but no experiences or memories. However, there is a capacity to learn. This can be likened to a new computer in that it too is mostly hardware with firmware to run basic logic check processes. Computers also come with the basic software to expand its capabilities.

The moment we enter the world, we begin remembering our experiences and reacting to them, whether the response is obvious or not. Early experiences are not recorded explicitly, with only basic characteristics such as pain, hunger, thirst, and other bodily needs. Over time we develop the capacity to actually remember significant events. At this point, we begin to remember all facets of an experience, including what led up to the event, what the event was, and was the response was. These first experiences are crucial, as they will be recorded, or remembered, when encountering similar events. This also explains why early childhood memories often have a significant impact, and why over time the impact may fade, as more memories (programs) are accumulated.  Generally, an experience or situation will be experienced countless times, such as walking down stairs, or eating an apple. After a certain situation has been experienced a certain number of times, an average memory program is created in our memory. When we need to do something, we simply recall this program and run it. Why is it that one does not always do something in precisely the same way? All programs will have a certain range of allowable error. Generally, the range is just enough to carry out the event while satisfying other factors.

If one was hungry, the simplified process might be as such:
- The person feels hungry. Depending on what the body is craving at that particular time, sugar, for example, a craving will emerge for sugar. This is similar to a hardware signal.
- Memory will be searched for closest available sources of simple sugars. Up until know, the total amount of time spent has probably been less than one second, with no real conscious effort. Once the target food is known, the basic "programs" necessary for obtaining it will be run. These could be walking to the pantry, opening a door, opening a wrapper, removing a cookie, and eating a cookie.
- These programs are executed in the correct order, all of which were unconscious efforts. After all, how many times have you consciously thought about something such as getting a cookie from the pantry to eat?

All of our actions are determined by our hardware signals and program signals. Our actions are determined by the most immediately significant task at hand. Generally, the hardware, genetic and basic needs, signals supersede the software signals in importance.

Differing personalities are simply different experiences and number of experiences. However, genes will influence which programs to run and how to run them. Some individuals are inherently violent, for example. For these people, their genes create a desire to execute programs that will cause them to engage in acts of violence.

The difference in consciousness between higher and lower level organisms is simply the number of body inputs, such as sight, feel, smell, etc. These will influence the complexity of programs that can be run.

I've gone off topic a bit, but overall I was trying to suggest that because of all the unconscious elements of "consciousness," all of it is in fact unconsciouses. In fact we are just sitting back while we see, feel, and hear ourselves running our own programs. Conscious organisms are just sitting in the audience watching themselves.

That made no sense.