A · Fool's · Wisdom

This is a song that I just finished writing, that in a lot of ways sums up the last year and a half or so. If you want me to play it for you, and you're somewhere where you can hear me, I'd be glad to.

I Believe

I crawl, I walk, I run, I fly,
I believe.
I laugh, I cry, I live, I die,
I believe.

I don't know how much more I can take;
My faith is tired, and my legs, they break.
I'm looking for justice, but only find grace.
What's wrong with me? Why can't I see my face?

I trip, I fall, I fail, I crawl,
I believe.
Your hand, comes down, to me, lifts me up,
I believe

I don't know how much more I can take;
My faith is tired, and my legs, they break.
I'm looking for justice, but only find grace.
What's wrong with me? Why can't I see my face?

I hear someone knocking at my door,
I'm ashamed to answer, so I try to ignore,
I look in the mirror to see "Who's there?"
I see nothing's face, no one's standing there.
I put my hand on the lock, and turn the key,
I'm frightened of You, and of what You'll see,
The door bursts open, I cry out in fear,
Your lips are smiling, Your eyes drip tears.
You take my hand, and You set me free,
You hold me tight, and You say to me...

"My son... That's all the faith you need...
That's all the faith you need!"

You hold me up, take away my weight,
Your love is strong, I can stand up straight,
You banished justice, You give me grace
Nothing's wrong with me; You can see my face.

I crawl, I walk, I run, I fly,
I believe.
I laugh, I cry, I live, I die,
I believe.

I love music.  I grew up with it, so I suppose it's not a huge surprise; but I love playing music, I love listening to music, and I very very rarely don't have music playing somewhere in my head, if not out loud as well.  So what is it about music that's so fascinating to me?  I know that when I'm playing, regardless of the circumstances, it's like a holy experience -- like yet another way for me to commune with God.  But it's also a way for me to communicate with others.  It's like a second language to me, one that everybody else understands, regardless of whether they're trained in music or not.  In every culture that I know of, music has played some part in their lifestyle, and it's usually a fairly prominent, tangible part. (I could be wrong, though, and I'd be interested in being pointed towards a culture in which this wasn't the case).

What is it about music that so completely pervades us as a species?  What defines some music as "good" and other music as "bad"?  I think part of it is the culture that one is raised in -- certainly, I, being raised in corporate America, am going to think that certain music is better than if I had lived in Japan several hundred years ago.  However, not having the experience of living in Japan several hundred years ago, everything further I say in this post is going to have to be taken as tainted with the corporate America bias.

I'm going to talk about several different styles of music, and what makes me think that these styles are good or bad.  Note that when I say "style" of music, I mean something slightly different than when I say "genre."  It's a very subtle difference, and I'm not even sure I can explain it in words, but I'll try.  I use genre to refer to an external description of music.  However, the styles of music refer more to an internal "mood" of the music.  A piece of music can fall into one particular genre, yet be in a completely different category in my mind.  I don't know if that makes sense to anyone besides me, so if it didn't make sense, just think of these as genres, because the difference is so subtle that in most cases, it won't make any difference.

• Classical: This is probably the starting point of modern music, in my mind (yea, you could probably argue about this with me if you wanted to, but just let me run with it), so it is interesting to me on several levels: first is a purely intellectual level.  It's interesting to see the beginnings of other music and discover where they come from.  (For instance, did you know that in some Led Zeppelin song they play the beginning of some piece written for classical guitar that I since forgot the name of?)  Classical music also just has a really intellectual structure to it, which appeals to me.  However, it also is frikin' gorgeous, and is fun to both play and listen to.  In some ways it's my favourite style of music, because it combines both intellectual stimuli and artistic beauty in one nifty package.
• Jazz:  By far the most expressive style of I've ever heard, which is only natural.  Jazz isn't about melody lines or tight structure -- in a sense, it's the exact opposite of classical.  The entire thing is focused towards on-the-fly improvisation.  However, it still maintains the same level of intellectuallicity (is that a word?) since in order to make good improvisation you must deeply and wholy understand the underlying chord structure, chord path, etc.  And it's a hell of a lot of fun to play.  I think that jazz far surpasses classical in some regards, because it's so spur-of-the-moment.  I think stuff created on the fly can be a lot more beautiful than stuff that's labored over for long periods of time.  Maybe.  The jury's still out on that one, but I think I like jazz better than classical
• Alternative:  This is your pretty typical "modern rock."  It's also by far the most common style of music that I listen to, which is kinda odd, given my love for jazz and classical.  It's a strange style of music that relies on tension and distortion to create something interesting.  Oddly enough, it succeeds.  I think it succeeds because it's interesting -- a lot of music in this style does new, unexpected, interesting things, which keeps me listening.  Some of it can get rather repetitive, but I tend to stay away from bands whose music all sounds the same, because it's just boring and sucks.
• Pop:  YECCH!!!  Gag me with a fork.  Please.  Popular music bugs the crap out of me, for lots of reasons, but the primary reason is that popular music is no longer about the music (I question if it ever was?).  It's all about image.  Britney Spears doesn't give a rat's ass if her music's good, as long as she's got the crowd worshipping her.  The music is just a stepping stone to get the crowds.  It no longer has meaning, which is a huge sacrilige.  As a result, we get music in this genre that sounds awful because the people playing it aren't musicians, they're performers.  It ends up all sounding the same, and generally is heavily digitally edited to cover up the fact that they can't play/sing worth crap.  That is all.
  
• Classic Rock:  Included in this are artists like the Beatles, Led Zeppelin, Simon and Garfunkel, Pink Floyd, etc.  These guys are really interesting, because there's all kinds of musical influences in their music, as well as a lot of new stuff.  They all have a lot of talent, they draw strongly from jazz and classical music, and they made it new because they were creative.  In some ways, they're the climax of music, because they were able to merge all kinds of music together, add a little bit of their own, and come out with something that sounds darned good!
  
• Country:  Country is.... odd.  A lot of it I dislike strongly, because it sounds too whiny: "My wife broke up with me cuz my girlfriend left me.  She backed over my dog on her way out and now I'm all alone.... Just me and my gun...."  (no actual country songs were harmed in the formation of those lyrics).  However, some of it I like, because the artists put a lot of thought and feeling into their words
  
• Rap:  I'm hesitant to even include rap under this list, because while it is a form of expression, it doesn't have any of the underlying harmonics that form music... I don't like rap much, though it makes me laugh sometimes.  That's all I'll say about that.
• Worship:  Worship music is completely different from any of the other categories that I've listed in that it serves one purpose, and one purpose only -- to aid the participant in worshipping God.  Therefore, if it succeeds at this task, it is good, and if it fails, it is bad.  A lot of worship music that I love I would otherwise consider "bad" music, because it is usually simple and repetitive.  Most worship music doesn't make suitable listening music because of this, and all worship music should not be slapped on a CD and labelled as such.  Many Christian artists would do well to take heed of this and not jump on the current "worship music yay!" bandwagon.
• Christian music:  This category can and does encompass all of the above categories, and in general what I said in that category applies to the Christian version of the music, with one notable exception:  a LOT of Christian artists SUCK.  They can't play worth jack shit.  Yet they've become extremely popular in Christian circles, and as far as I can tell this is only because they slapped the "Christian" label on their records.  News flash, people:  just because it's a "Christian" band doesn't entitle you to produce crap.  In fact, I would argue that we should be holding our Christian musicians to an even higher standard than we hold "secular" music.  We wouldn't put up with half the crap out there if it was in the secular culture.... Get a clue, people.   (sorry -- that was my little rant against Christian music).

Hmmm.... that was a really long post.  I think I've said everything I wanted to say... music fascinates me.

Hehe. That sounds like the title of a really bad movie... However, it's this really interesting concept that I just ran across tonight. First of all, go take a look at this, then come back.

Did you read it yet? Didn't think so.

Are you going to? Didn't think so.

So, the Uncanny Valley is this point at which things that are attempting to be human fail at being human, and instead just creep people out. The basic premise is thus: if you have a graph with the x-axis being how lifelike something is, and the y-axis being people's reactions to said object, the graph will start out with a pretty neutral reaction for purely mechanical-looking things; as the object becomes more and more lifelike, the human will react better and better to it, until it peaks at a point which isn't lifelike, but it's close. Then the graph will plummet, and there will be a hugely negative reaction to anything that's ALMOST human, but isn't. Finally, the graph will rise again at the point when the object is indistinguishable from a human. The place in which the graph plunges into a decidedly negative reaction is called the Uncanny Valley, because whatever resides there isn't quite human, but it's so close as to be really creepy.

That probably didn't make any sense, because I don't have a graph here, I just have words. So if you're confused, go read the article and then come back.

Isn't that weird? Why should something that's almost human but not quite be so incredibly creepy? Think about it, you know it's true. Movie characters that act a little less human (maybe they have a halting stride or they talk in a monotone voice or whatever) -- aren't they about 10 times more creepy than even a big, drooling monster? What's up with that? The human brain is expecting something that's human just like it, because it looks SO close! But then when it does something that's just slightly non-human, it gets really creepy really fast...

I don't know enough about human psychology to explain this, but I'll definitely be thinking about it more.

However, I also wonder if this is true for something that isn't physically human. For instance, suppose someone developed a program much like ALICE (see the link on my sidebar), only significantly better -- so good that it's almost impossible to tell the difference between how it and a human would respond. Is there an "Uncanny Valley" for this, too? Is it possible that it would be so close to human that its responses would creep us out? Especially if we didn't know we were interacting with a computer program and thought there was a real person on the other end of that conversation?

I think it's pretty likely, actually -- I think that there's going to be a point in AI research where someone develops a program that interacts really similarly to how humans react, but instead of making people like it, it's going to make people fear it, because it's unnatural. It will be in that netherworld of being *almost* human, but not quite. Hmmmmm....

That raises a really interesting question in and of itself: What is the maximum point of imitability of any robot/AI program? Chances are that we are not going to be able to exactly imitate humanity for quite a long time, if we ever can -- which means that we'll be stuck forever in that Uncanny Valley, unless we can figure out some way to get around it. I think there might be a significant amount of research that should be put into quantifying that Uncanny Valley -- how close can we get to making something human before it creeps people out?

Just as a side note, this concept is really interesting in the field of horror books/movies/etc. The best way (maybe) to really scare the crap out of someone is not with gore, big monsters, or anything like that: take something that's human, deform it *ever so slightly,* and I think you'll scare people more that way than anything else you could do.

This post has been brewing around for a while, but it came to a head today when I realized something: The several HUNDRED songs that I have purchased off of iTunes are encrypted, and I CANNOT play them unless I have an iPod or iTunes. Well, this is all fine and good, because I have an iPod right now. But... what happens when it breaks? I run linux, iTunes doesn't run on linux yet (and it may not ever, I don't know and don't want to waste the money I spent on those songs on that off chance) and then I've got a bunch of worthless 1's and 0's that don't do jack for me. Well, this is all fine and good, I just decided that I'll use iTunes to convert them all to mp3's... kind of a pain, but hey, at least then I know I'll be able to listen to them once my iPod decides to crap out on me. Oh, what's this? We're sorry. You can't covert that file type to an mp3, because we want to make sure you're not pirating that music. Or who knows what their reasoning is.... and frankly, who cares?

Welcome to the world of DRM -- Digital Rights Management, for the uninitiated. But what it really is, is control. Money. Power. Apple has decided that they want the ability to dictate, who, when, and where you listen to your music. Now, granted, Apple's done a pretty good job of not making this TOO stupid, but when it boils down to it, Apple has more control over music that I bought and paid for than I do. (For me at least, this is going to change soon -- I've found a program that'll break the encryption, and then I'm probably going to stop purchasing stuff off iTunes -- at least, I will not purchase things nearly as much as I used to). However, go read Slashdot and you'll see several articles a day dealing with various companies trying to incorporate DRM into their products. Intel is hardwiring DRM into their new processors. Microsoft is building it into their operating system. Etc., etc., etc. All of this is in the name of combating piracy. (Translation: all of this is so the greedy big corporations of America can dig their fingers even deeper into your pockets and squeeze for everything you've got).

Well, ok, perhaps that's a bit cynical. Let's assume for the sake of argument that THEY really *are* doing this in an effort to prevent piracy. First off, who ever said that we wanted to put up with preventative measures for piracy? So piracy (i.e., stealing intellectual property) is wrong. Why should the people who don't pirate stuff have to suffer so much at the hands of the evil bittorrent P2P terrorists? That's quite ridiculous -- punish the people who deserve it, not the people who don't. I, the consumer, DEMAND complete control over what I have purchased! Dammit, I will not put up with someone else telling me that I can't use the stuff I bought FAIRLY FROM THEM!!! I want the freedom to be able to install my copy of Microsoft Windows on however many computers I own! I want to be able to listen to music that I purchased whereever and whenever I want to! It's my bloody stuff, keep your hands off of it!

Secondly, who ever said piracy is wrong? Why should I not be allowed to share my music with my friends? Of course, this is a really hot topic, and I wrote an entire 10-page paper on it for my hum class last semester, but the basic conclusion I reached is that we as humans don't have the right to prevent someone from making a copy of some piece of work and distributing it freely. The only thing we have a right to is the "fruit of our labor," and by distributing music, we're not depriving anyone of anything. The only way we deprive them of the "fruit of their labor" is if we try to sell something that someone else put effort into -- in effect, we're stealing money from them then.

However, the whole "is piracy right or wrong" issue is a little bit on the sidelines for this topic -- I'd be happy to debate it with you at some other time if you want. However, I do not live expecting to have my possessions that I legally and rightfully own to be controlled by someone else. Stop it. NOW.

On the Subject of Faith

"'Then He isn't safe?' said Lucy.
'Safe?' said Mr Beaver; 'Don't you hear what Mrs Beaver tells you? Who said anything about safe? 'course He isn't safe. But He's good. He's the King, I tell you.'" ~CS Lewis, The Lion, the Witch, and the Wardrobe.

Why can't God be safe?

Certainly, from an intellectual point of view I can give you some trite answer about how God expects us to take risks and to go outside our comfort zones and blah blah blah. All well and good, unless you are actually faced with the possibility. I'm not saying that I am -- but I'm not saying that I'm not, either. What happens when God asks me to do something that's not "safe"? Because I know He will, and whether that time is now or not, I need to accept the possibility and be willing to accept it when that time comes. Instead of cowering in my room afraid to read my Bible for fear of what God might ask of me through His Word.

"I believe; help my unbelief!" ~The Gospel according to Mark, page 22 line 9
"How hard it will be for those who have riches to enter the kingdom of God!" ~The Gospel according to Mark, page 25 lines 5-6

What things is God asking me to give up for His sake? Why am I so afraid to enter His presence that I must distance myself to a purely intellectual level?

What if God asks me not to go to school next year? Could I do it? I don't think so.

I don't have enough faith to understand what God's doing. I don't have enough faith to be of any use. I just keep doubting myself, and decieving myself, and running away. I don't want to run anymore, God -- but I'm scared to death of what happens when You catch me.

Lord, I want to believe. Help my unbelief.

I learned something today. Life isn't about me trying to fix the world. It's about me trying to serve the world.

So much of the time I find myself trying to do it all myself. I want to solve everyone's problems, and I want to make the entire world better. But I can't make the entire world better. And as I've been wrestling with recently, I can't even make one person's life better. So then I start asking myself, "What's the point?" After all, if I can't help anyone, why bother?

But that's not the point. The point is being willing to serve, even at great cost to myself.

I've been spending a lot of time thinking about faith and religion recently, and one of the things that's really been getting to me is the stubbornness of fundamentalist Christianity. Go to almost any Christian church today. Mention that you play Dungeons and Dragons. Or read Harry Potter. Or any number of other things. Immediately, like a flock of vultures, concerned grannies and pastoral staff will descend on you in an attempt to save your soul from the fires of hell that are licking at your heels.

Now the funny thing is, I agree with a lot of stuff that's taught in fundamentalist Christianity. Not the DnD/Harry Potter thing -- that's just stupid. But fundamentalists will tell you to vote for Bush (I did). Fundamentalists will tell you to believe in Creation (um.... I sortof do). Fundamentalists will tell you that abortion is wrong (I think so too). Fundamentalists will tell you that if you drink, swear, or smoke, you're going to hell. You know what I say to that? Fuck you! I will swear as much as I damn well please! You could go on naming things that fundamentalist Christians say and believe, and a lot of them I'd agree with; some of them I wouldn't.

But who cares? It's not the point! Honestly, people, it DOESN'T MATTER. Some people have spent all their lives trying to convince other people that their specific grain of Christianity is THE RIGHT WAY! I have two questions for people like that: a) What makes you think that you're so special that God gave you all this insight that He neglected to give to anyone else? b) What makes you think that you're so special that you know everything about my life and are capable of sending me to hell? Just... no.

I hope I'm not being hypocritical in writing this. Maybe I am.

So, um.... where am I going with this?

What would the world look like if Christians stopped bickering among themselves for a change and actually did the whole "Show God's love to the world" thing for a change? What if we decided that being descendants of monkeys really isn't such a bad thing, John Kerry's a pretty decent guy, and that homosexuals are people too? What if we recognized that life is not about being right, rich, or famous, but that it's about being servants? What if we stopped viewing people as things that need to be converted, and started viewing them as PEOPLE -- people who have feelings, who hurt just like I do, who love just like I do, who cry just like I do, who laugh just like I do? What if we didn't worry about the little stuff?

What if we stop trying to change the world and rather start loving the world?

OK, here's what I've been trying to lead up to this whole time: this is all fine and good for me to say, but it doesn't mean jack if I don't attempt to live what I just said. So for people reading this: What can I do to serve you? I don't care if it's big or small -- if it's within my power and ability to do so, I will. I will not try to fix you. I will not try to change your life. I will not try to convert you. I will not try to mold your life to be a copy of mine. All I have to give is service, and that I give freely.

I'm serious. What can I do to serve you? Please let me know -- this is more important than anything else to me.

This is where I live. Come find me. Anytime is ok:

David Morrison
dmorrison@hmc.edu
x72017
909-374-5627
East Dorm 163

So last time I went off on one of the physical aspects of AI, this time I'm focusing more on one of the informational aspects of it.

So there's this little thing called pain that plays a pretty big part in our lives. It keeps us from doing stupid crap like sitting on a hot stove, cutting off random fingers for fun, etc. But what is pain? Or perhaps, more importantly, could we program something to feel pain? At the very most basic level, at least, pain is just a bunch of chemicals that semi-randomly interact with each other, triggering an electrical impulse to our brain, which transforms that impulse into something negative, such that we won't do whatever it was we just did again. That's all it is -- an electrical impulse. It's been translated by our brain, but the fact that it "feels" like something has no meaning -- it's just an artifice of the way our brain stores negative stimuli. All it *actually* is is an electrical impulse.

So people ask the question all the time, how would a computer "feel" something? How can we make it "feel"? What would it "feel" like to be a computer? The answer is actually rather silly, and kindof useless -- the question doesn't mean anything. "Feeling" is fake. Looking at the example of pain, it would be very very easy to program a computer to "feel" pain. Let's use a bit of prolog code, since I enjoy prolog:

this_hurts(hand_on_stove).
this_hurts(gunshot_wound).
this_hurts(chopped_off_finger).

I just defined three rules that would dictate what pain means to a computer. Obviously, for a human, the rules are a lot more basic (something like "this_hurts(chemicals_interacting).") but really, that's all our brain does. From this, it becomes trivial to write an inference rule that avoids things that produce negative stimuli:

action(X) :- \+this_hurts(X).

In other words, action X will only be done if the rule that says this_hurts does not exist. This is exactly what our brain does (on a slightly more basic scale, of course) when it experiences pain. It's merely a rule system, which is something that we can program into a computer, and it will be indistinguishable from our reaction. Granted, we might program the computer to respond differently than we might -- but still, it's just a rule that the computer reacts to. And this is what pain is.

This is easily extended to the entire range of human feelings: joy, sorrow, hatred, love -- we can cover all of them by a rules-based system like the one described above. So why hasn't a good program been written to emulate human feeling?

I think the answer partly stems from the fact that humans can irrationally choose to do something that the rules in their brain tell them is a bad idea. For example, we know that "this_hurts(gunshot_wound)." is true. However, we might choose to take a bullet for someone anyway. Why? This is completely illogical -- in fact, it breaks the "set-in-stone" rule that we've defined above, i.e., that we don't take an action unless there is no rule saying that it hurts. How can we program a computer to break rules? Or perhaps more importantly (because I can make a computer program that ignores every rule that is passed to it), how do we program a computer to *sometimes* break rules, and what sort of rule could we give it to determine *when* to break the rule, and when to uphold the rule.

The most immediate answer I can come up with is some form of a neural net -- in other words, have a network of rules that are weighted by some "threshold" frequency, and some action only follows if the threshold frequency is high enough. For example:

%% The .99 describes the threshold breach necessary to take such an action anyway.
this_hurts(gunshot_wound, .99).

%% A inference rule that says basically "If I do any action involving
%% some_girl, automatically breach the threshold.
love(some_girl, 1).

%% Check to see if the threshold potential of the action breaches the
%% threshold frequency of the consequences:
action(jump_in_front_of_bullet, some_girl) :- love(some_girl, X),
this_hurts(gunshot_wound, P), X > P.

Ok, so my syntax isn't very good, and this algorithm would need a lot of work before it could become useable, but the idea still is plausible. However, even this wouldn't be capable of expressing human emotion, because necessarily, we now need to be able to dynamically change the threshold frequencies at run-time (on the fly).

Hmmmmmm... So I'm basically just running a stream of consciousness right now, I'm not really sure how coherent or logically sound any of the above ideas are. I need to think more on this.

So I've been reading this book, it's called "How the Mind Works" or some such inane title like that. It's really quite an interesting book, and within the first few pages it got me thinking about sight. Humans are very very pattern-based people. The alphabet is full of patterns, there are patterns on the road when we drive, patterns in music -- basically everything we do is driven by patterns. So if we ever want to program a computer to be able to live intelligently in a human-dominated world, we're going to have to come up with some mechanism of making that computer interpret patterns. Consider the following:

Suppose we want to program a computer to recognize a stop sign. Simple task, no? On the contrary, this is a very very very difficult task. In order to recognize the sign, we have to have something to compare it against, whether it's another stop sign, or some other kind of description, either of which are entirely dependent on patterns (is it red? is it octagonal? does it have STOP in white letters written across it?) The easiest method (I think) would be to have a picture to compare it to, then we just have one simple comparison to make -- see if the two pictures are the same (to within some margin of error, of course). This is a fairly straightforward comparison. However, things get really really complicated really really fast. What happens if, for instance, the sign is rotated? What if it's not the same distance away? What if there's graffiti on the sign? Sure, you could have comparisons for each of these cases, but that's going to get out of hand really fast. You could easily have a million different comparisons, just for one stop sign -- which means you'd have to a) store a million pictures of a stop sign, and b) run a million tests to see if they're the same, within some margin of error. And that's just for ONE stop sign!

But maybe there's an easier way: what if, instead of doing a million comparisons, you had some sort of normalizing thing that takes the image and scales it to some normal size and orientation? Then you'd just have to do two operations -- a scaling operation and a comparison. Surely this seems reasonable, doesn't it? There's actually quite a few problems with this: first of all, how do you define what "normal" is? and how is the computer supposed to figure out whether what it's looking at needs to be normalized, or if it's a different object that's SUPPOSED to look like that?

I think that this issue is probably a little easier than it seems at first -- I mean, already we have face-recognition software and fingerprinting software that seems to work reasonably well. Sure, it's not as fast as we like, but we can make it faster, right?

Well, probably, but here's the REAL issue with this whole visualization/comparison thing for computers. We just went through all that work for one stop sign. Suppose now we want to add recognition capabilities for ALL road signs. Let's also suppose that we get the comparison time down to some outlandish number -- 3 billionths of a second. This means that our current processors can make 3 billion comparisons a second. There's probably about 50 different road signs, which really isn't that bad -- if we want to, we can look up 60 million road signs a second at that speed. However, there's a lot more in life than just road signs. Maybe we want to be able to recognize cars, too. There's probably a couple thousand different kinds of cars out there, so if we add that to our database, we can do 1.5 million comparisons per second. And then maybe we want to recognize different kinds of animals; let's be generous and only put 5000 different species in our database. That cuts our comparisons down to 425,000 comparisons per second. And if we want to be able to recognize common household items (lets say there's around 500 of them, including things like clothes, kitchen utensils, computer stuff, tvs, etc.), musical intstruments, office supplies, and on and on ad nauseum, you will very quickly bog down the computer to the point that it's unusable! And this is ONLY with doing comparisons, and doing comparisons insanely faster than is possible on any piece of equipment we have today!

Now add into this the fact that in every frame of video that a computer will be processing, there's easily 1000 (and I think that's being generous) things that the computer will have to compare. (Just look in some random direction, and count the number of individual, distinguishable items that you can see.) So say we have a database with 1 billion items in it, and lets say that you take in 60 frames a second. Not even addressing the problem of figuring out where one item stops and the next one begins, that's 1 trillion comparisons PER FRAME. Which means it takes modern processors 333 seconds to process one frame of video stream! (That's 5 minutes, btw). And it's getting 60 frames a second -- this is completely out of the question.

This raises a very interesting question: How the CRAP does our brain, which processes instructions SLOWER than modern computers, do it? It's completely and utterly insane. Figuring out how to make computers emulate our vision processing is going to be a significant barrier to making computers that are autonomous in a human world.

There are two things that I think might help with this problem, but I need to think about both of them more before I go anywhere with this: First off, our brain truly is multi-threaded. In other words, it's capable of actually doing lots of things at the same time. We don't have any processors that can do this. I'm sure once we figure out a way to do this, a lot of problems will become a lot easier. Secondly, given any frame of video, in general, things aren't going to change very much from that frame to the next one. Over time, yes, the image will change. But the computer isn't going to have to re-calculate everything for every frame. It ought to be able to just figure out what's new in the frame, and run comparisons on that, which would speed things up considerably.