decoration decoration
Stories

GROKLAW
When you want to know more...
decoration
For layout only
Home
Archives
Site Map
Search
About Groklaw
Awards
Legal Research
Timelines
ApplevSamsung
ApplevSamsung p.2
ArchiveExplorer
Autozone
Bilski
Cases
Cast: Lawyers
Comes v. MS
Contracts/Documents
Courts
DRM
Gordon v MS
GPL
Grokdoc
HTML How To
IPI v RH
IV v. Google
Legal Docs
Lodsys
MS Litigations
MSvB&N
News Picks
Novell v. MS
Novell-MS Deal
ODF/OOXML
OOXML Appeals
OraclevGoogle
Patents
ProjectMonterey
Psystar
Quote Database
Red Hat v SCO
Salus Book
SCEA v Hotz
SCO Appeals
SCO Bankruptcy
SCO Financials
SCO Overview
SCO v IBM
SCO v Novell
SCO:Soup2Nuts
SCOsource
Sean Daly
Software Patents
Switch to Linux
Transcripts
Unix Books
Your contributions keep Groklaw going.
To donate to Groklaw 2.0:

Groklaw Gear

Click here to send an email to the editor of this weblog.


Contact PJ

Click here to email PJ. You won't find me on Facebook Donate Paypal


User Functions

Username:

Password:

Don't have an account yet? Sign up as a New User

No Legal Advice

The information on Groklaw is not intended to constitute legal advice. While Mark is a lawyer and he has asked other lawyers and law students to contribute articles, all of these articles are offered to help educate, not to provide specific legal advice. They are not your lawyers.

Here's Groklaw's comments policy.


What's New

STORIES
No new stories

COMMENTS last 48 hrs
No new comments


Sponsors

Hosting:
hosted by ibiblio

On servers donated to ibiblio by AMD.

Webmaster
The computer does not understand the semantics | 756 comments | Create New Account
Comments belong to whoever posts them. Please notify us of inappropriate comments.
two nitpicks
Authored by: alisonken1 on Thursday, July 19 2012 @ 03:38 PM EDT
On number 2) - that's at a higher-level than the cpu architecture. Until such
time as AI becomes part of the chip hardware itself, it's still just a program
algorithm that must be executed by existing hardware technology - and that's
where he's describing hardware v. software making a "new machine".

---
- Ken -
import std_disclaimer.py
Registered Linux user^W^WJohn Doe #296561
Slackin' since 1993
http://www.slackware.com

[ Reply to This | Parent | # ]

Nitpick 1
Authored by: stegu on Thursday, July 19 2012 @ 04:17 PM EDT
The statement that "DRAM capacitors are not transistors
and charging them does not represent the creation of
an electrical pathway" (paraphrased, it is stated a few
times with slightly different wording in the text)
is formally somewhat incorrect.

The DRAM capacitors are literally the insulated gates
of MOS transistors, and charging or discharging them
does create and destroy (reversibly) an electrical path
between the source and drain of those transistors.
Same for flash ROM memory - insulated gates are charged
to make a MOS transistor conduct or block a current
between the source and drain terminals.
Therefore, A memory cell does indeed function very much
like a switch at the individual bit level, even though
it certainly does not magically re-wire the CPU to
perform something it was not supposed to be able
to do already by design.

The article is well written and largely spot on,
but this stands out as a point where it could be
attacked unfairly and far too easily. The wording on
this particular issue could be changed to guard against
such attacks. It is not a main point, but opponents
in the patent lobby will try to find the weak spots
and concentrate on them, because they probably have
no useful defense against the main points that are
made - other than pointing to old, stale and incorrect
case law that needs to be superseded.

[ Reply to This | Parent | # ]

two nitpicks
Authored by: PolR on Thursday, July 19 2012 @ 04:27 PM EDT
Granted, the paths aren't really "new" - they were all anticipated when the hardware was designed. That's where your argument should focus.
The paths were not "anticipated". They are physically etched on the chip at the factory. The transistors are varying voltages in paths which are always present. No new paths are created. The article states this point very clearly and multiple times.
It's possible, and increasingly being done in artificial intelligence, to implement semantics using an algorithmic system. Even your example betrays this: you say that a computer can "recognize" the letter F. Semantics aren't magic; if humans can master semantics, so can computers.
Computers mastering semantics is an illusion in the eyes of the observer. If you read the technical manual of a CPU you will find an exhaustive list of all the instructions which could ever be used in a program for this CPU. None of them allow the computer to process the meaning of the bits.

Recognizing a symbol such as the letter F is not recognizing its semantics. It is a syntactical operation. Computers are good at manipulating syntax.

Computers can only execute mathematical algorithms in the sense of computation theory. This is because executing the instruction cycle is the only thing a computer can do. A program, even an advanced artificial intelligence program, is input given to the instruction cycle. And mathematical algorithms are by definition mathematical entities which do not process the semantics of symbols. They only act on the raw uninterpreted symbols. This is part of the mathematical basis which makes the stored program architecture possible. You are disputing a fundamental limitation of the mathematics of computer science which is applicable to all programs.

A wisely chosen algorithm will manipulate syntax in a way that is consistent with semantics. Part of the art of the programmer is to find such algorithms. Artificial intelligence is no exception. When such a program runs the observer will have the impression that the computer understands semantics because it produces outputs which are consistent with a proper understanding of semantics. But the computer doesn't actually process the semantics. If we look at the internals of the program we can only find some pushing around of the raw uninterpreted bits. It can't be otherwise. Understanding of the semantics is not part of the CPU instruction set and it is not part of the mathematical computation theory which underlies the stored program architecture.

[ Reply to This | Parent | # ]

The computer does not understand the semantics
Authored by: cjk fossman on Thursday, July 19 2012 @ 09:31 PM EDT
The software "understands the semantics" to the extent that
the programmer makes it so.

[ Reply to This | Parent | # ]

Both utterly wrong.
Authored by: BitOBear on Thursday, July 19 2012 @ 11:17 PM EDT
Computers don't know or ever see an "F". That thing between those
quotes is the number seventy (70), a.k.a 0x46 in the ASCII and Unicode character
sets. Something in the -display- on a text-terminal would draw the picture of an
"F" when done old-school.

On your computer right now, before your very eyes, some library (like Cairo, or
your X display, or the windows graphical context library) has *sprayed* some
bits around in a frame buffer, some on, some off, at a nearly arbitrary spacing
due to the window and screen size issues, to variously spray-paint that Effness,
that very "F" onto your eyes.

That "F", that 70, could have started life as 'E'+1, a.k.a. 69+1=70.

The computer doesn't know from "F".

Artificial Intelligences are non-existent. There are currently what I call
"coded intelligences" in my Science Fiction writings, where rules like
"Cat" -> {"fuzzy", "warm", "sharp"}
and "fuzzy" -> increment(happiness), "warm" ->
increment(happiness), "sharp" -> decrement(happiness) ;; if
happiness greater than zero then "I like %noun%".

But the computer doesn't know what "cat" really means.

See how my logic "abstracted away" "cat" to happiness +1 +1
-1 ?

===

Similarly, you need to go re-think what "transistor as switch" really
means.

The presence of a valve is part of the definition of the machine. A machine with
valves has valves in particular places. Those valves can be on or off. The
-result- of the machine is changed by altering those valves. The machine isn't
rendered new.

Transistors are electronic valves. In fact, in England English, what Americans
call "tubes" (electronically, for being in glass tubes) the Brits
called valves (for describing their ability to block or allow the flow of
electrons). Transistors are "tubes" or "valves" in this
sense but without all the tedious glass, preheating and vacuum.

So in the sense that going into your bathroom and turning on the tap valve on
your sink doesn't give you a new house, having a transistor allow-or-disallow
the transfer of electromotive force (voltage) from the power supply rail to a
particular fragment of other circuitry doesn't cause device to be born anew each
time you change your mind about between "allow" and
"disallow".

[ Reply to This | Parent | # ]

Nitpick 1.
Authored by: Ian Al on Friday, July 20 2012 @ 07:03 AM EDT
The 'legal argument' is that the presence of programs switches paths between the
computer components. The charge in DRAM switches FETs on or off, but those FETs
do not switch paths.

There are alternative RAM technologies including static RAM which comprise, in
essence, transistor pairs in a bistable circuit such that when one is on, it
holds the other off. Again, flipping in either direction does not switch a path
between components.

The entire executing circuitry of the computer is made up of switching
transistors. The transistors are formed into boolean logic elements. The output
of the logic elements is dependent on the input. The logic elements do not
switch paths in the computer. Also, the interconnections between the logic
circuits are not changed by the program being in the memory nor by the
instructions they represent being executed by the universal algorithm.

---
Regards
Ian Al
Software Patents: It's the disclosed functions in the patent, stupid!

[ Reply to This | Parent | # ]

Groklaw © Copyright 2003-2013 Pamela Jones.
All trademarks and copyrights on this page are owned by their respective owners.
Comments are owned by the individual posters.

PJ's articles are licensed under a Creative Commons License. ( Details )