[FRIAM] What is an object?

uǝlƃ ☣ gepropella at gmail.com
Thu Jul 19 11:42:43 EDT 2018


There's also a deeper objection to this than Marcus makes, that of "data driven" modeling.  I fight this battle all the time at biological modeling conferences.  Most modelers *do* develop models based on ideas ... or, more technically, abstract hypotheses about abstract things (e.g. lipophilicity).  Their "best practices" suggest they can do that and then, later on, fill in those parameters/variables with numbers derived from actual measurements of experiments.  So, their conceptual models are then *enlightened* or guided by the real biology.  But the model's gist is still something very much like an "idea".

Data driven modeling takes a different approach.  It _attempts_ to derive models *directly* from the biology (as directly as possible, anyway), rather than going through us (obviously fallible) human abstraction machines.  Machine learning is an attempt at this.  Teh *-omics are attempts at this.  Etc.  And while it's (currently) true that such modeling efforts remain, in general, less efficient and effective at building useful models, they are making some progress.  E.g. https://www.news-medical.net/news/20180712/Study-suggests-database-analysis-better-predicts-toxicity-of-chemicals-than-animal-testing.aspx

So, no, it does NOT go without saying that one's ideas influence the programming.  It's true pretty much everywhere, but it does "go without saying".

Add to that my dead horse: that all ideas are actually biology, anyway, and it would be more accurate to say that a programmer's biology influences their programming.  And that's obviously true to any programmer who's put off going to the bathroom for just ONE MORE edit-compile-test iteration. 

On 07/19/2018 07:32 AM, Nick Thompson wrote:
> Well, it goes without saying, doesn’t it, that it’s your current IDEAS of biology that influence your programming, not biology itself, right?  And your biologiized ideas of programming then influence your notion of the cell.  We never really know clouds themselves.  So to speak. 

-- 
☣ uǝlƃ



More information about the Friam mailing list