Preferences

swannodette parent
I think most people on this thread are missing the point entirely. Computer Science is the sustained study of processes and how to encode them. Sussman is simply pointing out the most complex, powerful (yet flexible) processes that he is aware of and stating that our models of computer programming can't even begin to describe such processes.

Seems like just the kind of thing that Sussman, who has been invested in logic programming, and constraint logic programming for 30 years, would be interested in.

Sad I'm going to miss this talk, I'm becoming convinced that future programming languages will need to combine our rich knowledge of object oriented, functional, and (constraint) logic programming.


elviejo
Completely agree with you. In "My Favourite Interview Question"[1] The author asks: How would you design a Monopoly game? He goes on to say that with 'basíc' OOP you can model the elements: dice, buildings.

But What about the rules? One of his suggestions is to look to the Strategy, Visitor, and Command patterns.

But I disagree. I want to model the rules using Prolog! That is what prolog is great at.

So can please the next high level language standup? I just want: top of the line OOP (Smalltalk) constraint programming (prolog) functional (haskell) And Design By Contract (Eiffel)

And no... I'm not asking for the kitchen sink language [2]

Is simply that this concepts aren't exclusive and all of them helps us to better model reality.

[1]http://weblog.raganwald.com/2006/06/my-favourite-interview-q... [2]http://zedshaw.com/essays/kitchensink.html

jodrellblank
Isn't this something the .Net Framework should be good at? Integrate your OOP (C#) program with functional (F#) modules, and so on. There are implementations of Eiffel, Smalltalk and Prolog for .Net too.
Confusion
Well, it's a bit of an attention seeking title and it seems people are taking it literally. I wouldn't bet that Sussman literally means it. "We really don't know how biology computes" sounds more like it.

I have never seen any reason to assume that there's something deeply different about the way the genome or the brain performs computations. My guess is it will turn out to be interesting from a technological point of view only.

babel17
It always astounds me how people can completely blend out one major thing: consciousness

Is it possible that machines develop consciousness like ours? THAT is the question that needs to be answered and which is far more interesting than "from a technological point of view only"

Confusion
To me that's not a question at all. The answer is a ringing: yes, of course. I'm squarely in the 'consciousness is an emergent phenomenon of certain configurations of matter', so to me it is only a question of the technology to create those configurations.

There is matter and only matter. Humans are a configurations of matter and those specific configurations have the property we hold so dear and call 'consciousness'. There is no reason whatsoever to suppose you couldn't synthetically arrive at a configuration of matter that displays the same emergent behavior. This is the basic thesis of the famous book 'Godel, Escher, Bach' by Hofstadter. I am a strange loop. Any similar strange loop will display the same properties.

Any question to this regard probably makes you a closet Cartesian dualist. When you drill down, most people turn out to be that. (Some intermediate philosophical positions are possible, but they are subtle and rarely held consistently by someone that hasn't spent a course studying the matter in detail).

What I'm guessing in my previous post is that it may turn out that you can most easily achieve a configuration that displays consciousness by using biological materials. In that case, any future conscious machine would probably have a core of material similar to our brain. Which more poignantly raises the ethical questions related to creating such machines.

(BTW, note we are conflating 'any consciousness' and 'intelligent consciousness' by focusing on humans)

This item has no comments currently.