 |
 |
Hi Carlos, I think I was pretty clear. Pretty. Maybe not 100%. In a previous mail I also talked about this and your message surprised me. If you love computers, go with them. I don't see a reason to go further. Everyone has his/her beliefs. Best wishes, Lubomir -----Original Message----- From: PhD-Design - This list is for discussion of PhD studies and related research in Design [mailto:[log in to unmask]] On Behalf Of Carlos Pires Sent: Monday, January 19, 2015 8:53 PM To: [log in to unmask] Subject: Re: Automated image rhetoric and user characteristics assessment Hi Lubomir, Thank you for your reply. A few comments to your message and one request for clarification follow. On 19/01/2015, at 23:54, Lubomir Savov Popov wrote: > Hi Carlos, > > I am happy that you are such a great optimist about modeling. It depend how you interpret the concept of model. I don't know anyone up to now who have managed to model the work of human brain in a reasonable way. I think the concept of "model" is stable and consensual, and leaves no room for interpretation whatsoever. It is the depth and precision of the model that may vary, according to one's aim. Of course you don't know anyone who has managed to model "the work of human brain". That's a pretty vague statement that can mean anything and nothing, ergo it is meaningless. On the other hand, I was not clear in expressing that I was not thinking about a model of the human brain, but about models of design(ing), which is a whole different class of problems. > Most of the models are mechanistic and stochastic. The problem is not with computer power. This is a common misconception. The main advantage of the human brain is actually sheer, massively parallel, computing power: "The average human brain packs a hundred billion or so neurons-connected by a quadrillion (1015) constantly changing synapses-into a space the size of a cantaloupe. It consumes a paltry 20 watts, much less than a typical incandescent lightbulb. But simulating this mess of wetware with traditional digital circuits would require a supercomputer that's a good 1000 times as powerful as the best ones we have available today. And we'd need the output of an entire nuclear power plant to run it." Just take a moment to let that sink in. You would need 1000 top-of-the-line supercomputers and a nuclear power plant just to get the hardware to match the potential of 1 human brain. http://spectrum.ieee.org/computing/hardware/lowpower-chips-to-model-a-billion-neurons > The problem is with the models that are flawed and always one-sided. At least by my criterial. I don't know a single social model (except for models of mass processes) that doesn't have a myriad of flows. I don't understand. Are you saying that models are flawed because they are one-sided, or because they have a myriad of flows? > There are ways to computerize many of our activities. However, the success of such software will depend on the social models. Garbage in, garbage out. This is well known in sociology. And again, it depends what work we want to computerize. Spec writing? If we add numbers, computers can do that much better than us. Soon computers will drive cars better than humans. Actually, commuters fly cruise missiles since the 1960's and they do that better than any pilot. But as we see, here is no progress in computerized design. No (social) models, no progress. Forget about computers. They are the easiest part. Yes, there is no magic and no miracles. Garbage will be garbage. Yes, of course the most important aspect of designing a "computerized designer" is... to design it! I do believe it is possible and not that far. We shall see. Best regards, ================================== Carlos Pires [log in to unmask] [log in to unmask] ------------------------------------------------------------- Design & New Media MFA // Communication Design PhD Student @ FBA-UL Check the project blog: http://thegolemproject.com ----------------------------------------------------------------- PhD-Design mailing list <[log in to unmask]> Discussion of PhD studies and related research in Design Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design ----------------------------------------------------------------- ----------------------------------------------------------------- PhD-Design mailing list <[log in to unmask]> Discussion of PhD studies and related research in Design Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design -----------------------------------------------------------------