rebelscience.org — We are not used to think of it as such but a computer program is, in reality, a communication system. Problem is, in an algorithm, communication is limited to only two objects at a time. Computing will not reach its true potential unless we abandon the algorithmic model and embrace a signal-based, synchronous model.
Nov 21, 2005 View in Crawl 4
lazerf4rtNov 22, 2005
I skimmed this article looking for the "How We Can Fix It" part, and just found a lot of complaining, hand-waving, and finally (after following some links) a screenshot of some crappy program with a graph of an OR gate. I'm a programmer so I'm interested in ways of doing things, but this article stinks. This guy should do himself a favor, go get a job flipping burgers and stop torturing himself trying to reinvent computers.
Closed AccountNov 22, 2005
I can think of one reason that a connectionist model will fail outside of specialized applications: Users don't have the time or understand it takes to train a parallel, connectionist based system (and don't give me that tripe about Microsoft training it for them... every large data set problem is unique no matter how similar the tasks might look... this is why handwriting/voice recognition systems are still niche markets... duh).
pvaculinNov 22, 2005
The article gives a perspective for synchronous signal based computing. It is meant to give a background that supports the author's theoretical model. If one looks at the web page in which this article is found you would see that the basis of the web page is the development and funding of the COSA Operating System. The comments act as if this is fact. Like most scientific inquiry, a theory is postulated with reasoning. There is no quick fix. It is not until a person of this caliber gets to the peer reviewed stage that one could offer fact based criticism.
roy182Nov 22, 2005
This article is the biggest PIECE OF s**t ever captured by my brain - synchronously. The author is so clueless that is not worth disecting... "ICs are synchronous" f&^#!, "Event dependency".. Hey Eistein, there is something called the Observer Pattern.If you build s**tty state machines for your "Silver Bullet" you still have a synchronous piece of s**t. What an assh**e.
blitzennNov 23, 2005
What a bunch of crap! His statement to start his whole article is severely flawed. That is that computers, as they exist today, cannot be adapted to real world simulations because it cannot do computation in a parallel fashion needed to do so. Wow! Does he every have his head in the sand. Parallel computing tasks are the bastion of many a supercomputer, are there are many of them these days.
jobeemarNov 23, 2005
"Synchronous software will become a requirement as the hardware becomes more parallel."This system described on the site has little to do with hardware parallelism, in my opinion. It has to synchronous reactive software wich can be simulated on a single CPU. Its main purpose is to solve the reliability problem, not the speed problem. Having said that, I agree with you that reactive systems are perfectly suited for parallelism. At least, that's the way I understand it.
ringo47Nov 23, 2005
To all the people who said "Looks interesting, I'll read this later.": Don't waste your time.From the abstract: "In this article, I will propose a silver bullet solution to the software reliability and productivity crisis."Fred Brooks: "Not only are there no silver bullets now in view, the very nature of software makes it unlikely that there will be any..."I'm not judging who's right or wrong, but it's a little hard to take statements like this seriously:"Most people in the software engineering community wrongly assume that algorithmic software is the only possible type of software. Non-algorithmic or synchronous reactive software is similar to the signal-based model used in electronic circuits. It is, by its very nature, extremely stable and much easier to manage."And that's about as much detail he gives about his silver bullet. The rest of the article is spent extolling "non-algorithmic" software, never explaining what that means, and vaguely implying that software should be more like hardware - just without all those pesky algorithms to making it unreliable.