[Taxacom] Automatic species identification: DNA only descriptions

[Jason Mate]

>Alas in Taxacom there has been a tendency to identify these efforts as DNA-based, when this assertion is clearly a fallacy


>>yes, but the other attempts at automated ID have clearly failed, leaving DNA as the only remaining "great white hope"


>>and when nobody but machines know how to do IDs, and they go wrong (as they will), who will know?

Look, I don´t know what to tell you. We are either going to be out of a job because machines will do it faster and cheaper than us or because there will be nothing left to identify. ;b

Jason
 


From: Jason Mate <jfmate at hotmail.com>
To: Taxacom <taxacom at mailman.nhm.ku.edu>
Sent: Mon, 23 August, 2010 7:33:39 AM
Subject: Re: [Taxacom] Automatic species identification: DNA only descriptions


> > Whilst at the BMNH I remember at least 2 sepparate projects to automate species identification. Both used morphology. The dream to automate identification is pervasive and not the domain of one or another group. Whilst it might be entertaining to identify organisms it is a practical matter of workload management. "Never have so many asked so few to identify so much"
> 
> The current hype on automatic species identification circles around
> two myths: that it should be easier to build an automatic species
>
 identification system if DNA is used for the task, and that this
> identifier will be error-free.


Yes, but the point I was making is that regardless of the data-source you trust/believe-in, efforts, misguided or not, are afoot in accelerating identification. Alas in Taxacom there has been a tendency to identify these efforts as DNA-based, when this assertion is clearly a fallacy.


> 
> Among the many obvious problems:
> 
> 1. DNA makes thing easier sometimes and much more complicated some other times.
> 
> 2. Any identifier (even a human one) will be error prone.
> 
> 3. The decision on which method should be used for the assignment and
> how to deal with sensitivity to method choice is... a big issue.
> 
> The failure of most automatic identification systems seems to be that
> many people is hoping to have some kind of taxonomist droid capable of
>
 classifying whatever comes across... this hope is naive at best.
> Automatic classification is not an easy task. There are plenty of
> problems in automatic classification (whether supervised or
> unsupervised) that we biologists seem to ignore, I guess this is the
> root of over-optimism regarding automatic taxonomic classification.

We are moving away from the topic but I will humour you. The problems you have listed are applicable to pretty much anything just by substituting DNA and identifier for other nouns. Hence automatic identification is like any other activity you can imagine. That seems like an endorsement.

Now, is there an android in the horizon that can identify organisms like humans can? No. Will it ever happen? Yes, eventually the computing power of a laptop will match and then surpass us. Of course this will probably not affect you or I but it is entertaining to think of cybernetic taxonomists doing
 the donkey-work. At least the sorting :)

Best 


Jason
                        
_______________________________________________

Taxacom Mailing List
Taxacom at mailman.nhm.ku.edu
http://mailman.nhm.ku.edu/mailman/listinfo/taxacom

The Taxacom archive going back to 1992 may be searched with either of these methods:

(1) http://taxacom.markmail.org

Or (2) a Google search specified as:  site:mailman.nhm.ku.edu/pipermail/taxacom  your search terms here