Skepticism and Certainty

[TINNY]

To a skeptic, is there such thing as Certainty? To be skeptical of everything, I don't think so. So how can skepticism be pragmatic? Is logic perhaps what they call certainty? What if logic has a priori? Certainty implies no a priori right?

 

Descartes had this to say about certainty: "I think, therefore I am".

 

However, this has been refuted by Bertrand Russell, quite convincingly at that.

 

(I'm on a thread-starting rampage)

[Buffy]

I've always prefered the variation of Descartes: "I think that I think, therefore I think that I am...."

 

Uncertainty is not by definition bad unless it leads to paralysis. Mr. Spock would always argue alternatives by stating the probability. I don't really see anything wrong with basing ones actions on most likely outcomes, and I think its a lot healthier mentally if one is prepared for the possibility of other outcomes rather than being devistated by finding that one's fervent and strongly held beliefs are not true. :hyper:

 

Logic on the other hand, is logic, and it will properly lead to conclusions if the assumptions are true (sometimes that's a BIG if!). Probabilistic or "Fuzzy" Logic (thanks Lofti!) is an interesting merger of these two concepts (simple: http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol2/jp6/article2.html not so simple: http://www.generation5.org/content/1999/fuzzyintro.asp

 

Cheers,

Buffy

[alxian]

fuzzy logic is all well and dandy but does it rely on terabytes of stored experience or a few hundred thousand lines of tentative code?

 

i'd play chess agaisnt a machine that could make decisions i didn't expect... fuzzy logic = gimmick

 

i'd place my life in the hands of a machine that i knew knew what it was doing based on experience (not only its own operational experience but procedural knowledge stored locally (the machine cerebellum if you will, where the internet would be its equivelant of our two big lobes, and that each machine would only be a node on a network of thousands of machines who's idle cycles would be used to crunch bigger tasks instead of going to waste), and intuitive logic (what comes after fuzzy.. hopefully), which deals with stored knowledge and takes into account current events to MAKE decisions on the fly (but lots of data crunched and recrunched in parallel (like that base 20bit amino acid computer processing at 1024 bits per cycle at 20ghz per second (though speed won't be a factor only crunching numbers (since the human brain hardly ticks at all really).. but i'm just throwing numbers out there, when i should say that instead of meaningless bits future machines will have to deal with thought bits, entire files at once, uncompressed pure information). giving machines the human ability to process events faster than realtime, before they happen... (but not necessarily precognition, but not excluding it either)

 

in the future i hope that such machine will replace humans in some tasks that require faster procedural response than our flimsy nervous system and tennis ball knot of 'real brain' can muster.

 

i.e. when planes get too fast, cars get too fast, trains boats etc, it'll be nice to know that your machine (not necessarily anthropomorphic.. read: the charriot isn't just horseless its driveless, the driver being a omnipotent evermind crayX smart server hidding 2 miles underground operating its thousands of buses like cards hive queen (well not through telepathy, but through a secure wifi network) driver/pilot is faster stronger and smarter than a human would have been in that given situation.

[maddog]

alxian,

 

On who experience ? Not the machine it doesn't think. It doesn't experience. Cognitive thinking

creatures (Humans) do that. You can program a machine to "think". However, it is only a simulation and

not thinking ... yet. Fuzzy Logic can make a washing machine record a history of the clothes it has washed

and make adjustments accordingly. Does not imply that given a new circumstance of unknown origin

(strange piece of clothes never before seen) would this washing machine know for sure what action to take.

 

Maddog

[Buffy]

fuzzy logic is all well and dandy but does it rely on terabytes of stored experience or a few hundred thousand lines of tentative code?

 

i'd play chess agaisnt a machine that could make decisions i didn't expect... fuzzy logic = gimmick

 

Whoa! This is getting majorly off-topic, but just a point on this tangent: Fuzzy Logic actually does not require learning a la neural networks (it can of course, but if it doesn't, it doesn't need much of a database and the code can be pretty darn generic). Its simply code that combines data to come up with values that are matched against a set of probabilities that allow the code to go different directions based on gradually varying inputs instead of the good ol 0/1 binary used in formal predicate logic. And as Maddog points out, its not a gimmick, its likely in your washing machine, varying the water level and the spin speed based on the weight and quality of the clothes you put in it. As Maddog also points out, it won't deal with situations you didn't program into it, so you can't expect the washing machine to magically figure out what to do if you throw your car's carborator in it. Pretty cool stuff though and being used in LOTS of places....

 

Cheers,

Buffy

[Freethinker]

To a skeptic, is there such thing as Certainty? To be skeptical of everything, I don't think so.

Your confusing a Skeptic with a Cynic. A cynic will never accept anything as good enough proof. While a Skeptic just has well defined parameters which are acceptable as proof. A Skpetic just filters data for acceptable levels of logic and reason. A cynic just rejects it out of hand.

Is logic perhaps what they call certainty?

Logic is a tool used to evaluate information. It is used at two levels. One to evaluate the validity of the construction of the argument, does it pass simple process analysis. Quality of source, level of evidence, absence of simple fallacies, ...

 

Then it is used as the process of choice for combinging the seperate elements into a complete picture.

 

Any failure along the way is a NO. Lacking any failure, it is a YES, or to use your word, a Certainty. In fact you may wish to read "The Atheist's Certainty"

 

http://www.infidels.org/library/modern/peter_wilson/certainty.html

 

It discusses the idea of flying reindeer and how one might go about determining if it is true or false and to what level of "Certainty".

What if logic has a priori?

"Logic" can't. The data it is being used to evaluate CAN.

[alxian]

maddog,

 

It doesn't experience

 

of course a machine experiences. all it records is raw experience.

 

given that a machine can have senses that touch on eevry part of the electromagnetic spetrum a robots life would be far richer than a humans.. given unlimited memory capacity (and access times)

 

i believe you are mistaking my speach.. i'm trying to say modern machines are childs play things compared to what would be necessary for machine to rival human intelligence. (which i'm trying to also say is more hive-like than individual. a machine will suck.. many machines as nodes in a massive low latency high memory network will prove far superioir to humans.. maybe not individually but as a collective. they will have perfect memory (which humans rely on computers just to have a base amount of reliable memory (not even close to perfect) not to mention such a network will be able to while idle theorize potentials as humans do. we do this better as individuals (but sciences are best advanced with many minds than with only one (generally)) in this way humans and machines will be similar.. given that similarity and proximity i don't see why coexistance is such a hard concept. as long as we don't directly compete for any scarce resource and have a common goal (although what a machine's goal would be is undertermined, i'm sure it won't be perfecting or(by) inslaving humanity.. that is to say unless like the matrix machines figure out that indeed humans will be a means to not only our own destruction but theirs as well.. i think it would be gods will for us to be relieved of free will long enough for us to miss it and realise it should be cherised not exploited) like exploring the solar system and populationg it with a mechanohumanoid civilization (note i didn't say hybrid race)

[Freethinker]

fuzzy logic is all well and dandy but does it rely on terabytes of stored experience or a few hundred thousand lines of tentative code?

 

i'd play chess agaisnt a machine that could make decisions i didn't expect... fuzzy logic = gimmick

 

i'd place my life in the hands of a machine that i knew knew what it was doing based on experience (not only its own operational experience but procedural knowledge stored locally... (but lots of data crunched and recrunched in parallel... when i should say that instead of meaningless bits future machines will have to deal with thought bits, entire files at once, uncompressed pure information). giving machines the human ability to process events faster than realtime, before they happen... (but not necessarily precognition, but not excluding it either)

What you are talking about is not Fuzzy logic, but Expert Systems or Artificial Intellegence. The human mind is not a linear processor like our PC's have. It is more of a massive parrallel. Each a relatively small one.

 

As to a different type of processing, what you describe is called object-oriented Symbolic Processing or "List Processing". Where words or even "thoughts", strings of words, are disected and processed.

 

There is a computer operating environment called LisP (List Processing). There was a company that was started as a brain trust out of MIT's AI lab and heavliy funded by DOD/ARPA. It made a computer than ran native in a LisP environment. The company was Symbolics. Now you can get a LisP shell that runs in "C" and ported to various unix OS'. Neural Network software tends to use it.

 

http://smbx.org/

 

We used to explain it using American Express as an example. When you would charge using Master Card or Visa, the system would do a comparison of available credit to purchase amount. Yes/ No. But Amx did not have a Credit Limit. So a system had to be designed that compared charge types. If a charge request came thru for a Harley and the guy usually only used it to rent cars and Hotels, .... So an Expert System had to be designed.

 

To give some idea of how influential this company was, the first domain name ever issued was symbolics.com. The vast majority of computers on ARPAnet were Symbolics.

[Turtle]

Most of the original question/statement is slippery as an eel. But I can get a hold on ol' Renee. He should have stuck to analytic geometry & this is why. The idea of his cogito was to throw out everything he couldn't be sure of & then rebuild on a firm foundation of certainty. His error was the only thing you/me/he can be certain of is "I"; not cogito, not ergo, not sum. Ergo, he built on sand afterall. QED :hyper:

[Buffy]

What you are talking about is not Fuzzy logic, but Expert Systems or Artificial Intellegence. The human mind is not a linear processor like our PC's have. It is more of a massive parrallel. Each a relatively small one..

Totally! Fuzzy logic was invented by Lofti Zadeh, and has the def above...I got to learn it directly from him! So cool...

 

As to a different type of processing, what you describe is called object-oriented Symbolic Processing or "List Processing". Where words or even "thoughts", strings of words, are disected and processed.

 

There is a computer operating environment called LisP (List Processing).

There was a company that was started as a brain trust out of MIT's AI lab and heavliy funded by DOD/ARPA. It made a computer than ran native in a LisP environment. The company was Symbolics. Now you can get a LisP shell that runs in "C" and ported to various unix OS'. Neural Network software tends to use it.

I took a compiler class from a guy (Dick Fateman) who was into AI, and he evilly made us write a C compiler in Lisp. It was lots of fun, but Lisp is NOT for the faint of heart. Dedicated Lisp OS's are much more efficient than Lisp interpreters that run on standard operating systems because of the heavy need for garbage collection of memory, but that can be optimized when compiled. Lisp was originally designed as a purely interpetive environment and in addition to the ability to manage lists (great for organizing "concepts"), Lisp is the ultimate environment for building self modifying code: a Lisp program is a list which can run itself! If you're interested, I biasedly recommend Franz Lisp (http://www.franz.com, no commercial affiliation, just friends of the founders and a long-time user).

 

I apologize for the digression. Now back to our regularly scheduled debate.

 

Cheers,

Buffy

[Turtle]

The problem with getting detailed about this logic part, is that it is impossible to do the subject justice with anything short of the length of Dougy's "Metamagical Themas" or the depth of his "Goodel, Escher, Bach". I am certain of this. :hyper:

[Freethinker]

Totally!

Cheers,

Buffy

And lives in the Valley no less!

 

Gag me with a spoon!

[Freethinker]

I took a compiler class from a guy (Dick Fateman) who was into AI, and he evilly made us write a C compiler in Lisp.

Just for accuracy it is LisP (upper case P).

 

While Lisp is a speech impediment.

[Buffy]

Just for accuracy it is LisP (upper case P).

While Lisp is a speech impediment.

FT is right and everyone else is obviously wrong: Try Googling "common lisp"...

 

Gotta say, I've never seen it with that big P (although sometimes with a shouting, fortran/cobol-like all-uppercase rendering), but I have worked with a bunch of implementations over the years (as an exercise, run: "(setq foo 'L 'i 's 'P) (setq ft (cons((car foo), (cadr foo), (caddr foo), 'p]"!). Its probably one of those strange MIT affectations I'm sure. You know, bad water in the Charles River and all. :hyper:

 

Cheers,

Buffy

[Freethinker]

FT is right and everyone else is obviously wrong: Try Googling "common lisp"...

 

Its probably one of those strange MIT affectations I'm sure. You know, bad water in the Charles River and all. :hyper:

And you are probably right on the money. As an ex-Symbolian (and a member of the Symbolics "Usual Suspects") the capital P thing was drilled into my head. Yet I find it seldom used anymore. Erg.

 

But then every trip to our Cambridge HQ was a "strange MIT affectations" experience.