16th September 2013, 07:13 PM
P Prentice Wrote:hi doug - the intro in the phd paper sez it all really. the models you build are only as good as the data used to create them and this is all flawed or just wrong. there are too many variables and inconsistent ways of interpretting archaeology. i cant think of any model for predicting archaeological sites even using broad categories that has stood the test of even a small amount of time (in the uk and the dutch have probably lost more than they will ever admit by using crap pm). there is no basis for building an accurate model, there is not an smr or her that contains accurate information and certainly not one that you can easily quizz for accurate information, the speed of investigation and chance recovery over the past 40 years has meant that nobody has a clue as to what was lost without record and how that skews your models and there has never been a publication in any journal ever that has been able to show that the data used was accurate or nowadays, meaningful. the prospect of governments latching onto the pm heresy and thereby having statistisions evaluating development sites i find amusing indeed
Yeah- read on to the paper I posted about new developments in pm. You can make completely behavior based models that are really great in areas with no previous archaeological work (or as you point out all sorts of problems with the data). Using stats and previously known sites is only one method. I personally don't like it too much but it works. I use different methods
Also read my comments again, we have gone into a straw man argument here. I stated that even with really good models you always need to check. There are really good models out there but even if there are right 95% of the time they are still not perfect.
The comment that started this all was- we need to dig less because we don't find stuff. To which people said ah but sometimes you do. At which point I added using PM you can get really good results (not 100%) AND you should always check. The example I give in that paper is the perfect example of using PM and still needing to check. I never said PM was perfect. I said that in some cases- using PM we can be really dead on but still need to check. That is why PM is best used for desk based assessment not a replacement for anything. Pretty much everyone in the geopyhs discussion I think agrees that it is a tool to help but not a replacement and the same for PM. It was simply an observation that was relevant to a discussion about the fact we know sites are and are not certain places.
Now I know as a curator you might feel like I am stepping on your toes with PM. Depending on how you do PM it is exactly what you do- look, guess (educated guess) and recommend to planning that they should or should not do archaeology. How many sites have you lost by not requiring archaeology investigations on everything? Or do you recommend archaeology for everything? At which point please share with us how you keep your job and not piss people off?
I am not trying to replace your job. I am only adding another tool which may or may not work all the time- like geophys. A model still needs someone to interpret and make decisions with it. That also means looking at if the data is bad or there are problems with it etc. etc. everything you mention are things a good modeler should be concerned about.
If you want more discussion about problems with PM and the solutions I will be more than happy to point you in the direction of some resources. However, the points you bring up are complex and I don't really have time- nor would anyone want me to spend hours typing in 1000s of words on the subject here on this forum.