# Grand Challenges for Archaeology: a model based point of view

TL;DR – Current American archaeology lacks a cohesive framework for the construction, evaluation, and implementation of quantitative models specific to our domain…
and its my generation’s fault.

This post is in response to Doug Rocks-Macqueen’s  call for Blog Carnival to explore the Grand Challenges for Archaeology; an amazing idea! This call for blogs can be found at Doug’s blog, Doug’s Archaeology.  The idea is to collect a series of blog posts (or any web-based medium) based on this theme.

The source of this topic is a survey of archaeologists to determine the 25 grand challenges that we and our field face.  The responses from this survey (Kintigh were synthesized into an American Antiquity article Kintigh et al. 2014 (article available here and responses available here [login required]). I read this article upon publication and had a range of responses; from inspired to angered to critical to amazed.  Surely the signs of good thought provoking article.

“in God we trust. All others bring data.” ~ W. Edwards Deming

While the statement at the top of this post elicits a similar wide-ranging response in my mind, the thesis remains the same. This contention is developed based on 20 years of focusing on GIS, spatial, and quantitative analysis in a predominantly CRM context. I am guilty of every shortcoming, failure, and n∅∅b mistake one could endure while pursuing this knowledge over the years and it is through these setbacks that this point of view was developed.  At every inflection point I was sure that the there must be a body of literature just out of reach that would illuminate my questions ; surely somebody had addressed these issue already.  The truth became evident in time that a quantitative approach to archaeology was not contained within the neatly wrapped package that I had hoped.  Instead it is found evolving in a mosaic of niches, feeding off the remains of that which came before it. unconformably underlying this landscape is a rich stratigraphy of quantitative intentions gone by; some as diamonds and others as a dark smudge.

To unify and develop the quantitative archaeological landscape to be accessible to all interested archaeologists, regardless of their training, is to solve a grand challenge facing archaeology today.  The tools exist, the data exists, and the desire exists to do this, we just need the structural support on which to build it.

$\pi(\theta\mid y)\propto f(y\mid\theta) p(\theta)$

By “model based” archaeology I am referring to an approach that fosters the formalization of logical implications and consequences of observed data and theory.  Furthermore, formalization is carried out through quantitative methods that make explicit assumptions and results are reported in the context of the model.  Finally, this approach is supported by modern computational statistical methods and data visualization techniques that allow for effective communication to non-specialists and the publics.

The basis of this approach is not remotely revolutionary and it seems to me to be at the core of archaeological thought.  Taking noisy, fragmentary, and incomplete observations and combining them with previous (mental) models of a phenomena to derive a new understanding as a weighted average of all of our knowledge on the topic is pretty much what we do.  However, I believe we can evolve our collective instincts for building knowledge from data into a more unified, accessible, and robust approach.  Again, not necessarily a revolution (e.g. New Archaeology and The Model-based Archaeology of Socionatural Systems), but a call-to-arms to unify and bring our collective resources together for the establishment of this perspective as a more prominent feature of American archaeological discourse.  The tools, ability, and smart folks are all present, as evident by a number of people doing great work in this direction.  Let’s transform archaeology into a field where data is shared between sub-fields, models are developed based on problems specific to archaeology, findings are communicated with narrow and broad implications, and all archaeologists can access the tools to participate.

Now on to the assertion of blame… Truthfully, I am not into blame; what is is and what’s done is done. However, this is one situation where understanding the point of dissidence may be helpful in recognizing the problem more fully.  IMHO, is appears that my “Gen-X” cohort of archaeologists (myself fully included) failed to hold up our end of the bargain. Duly noting that it would be an unfair and gross overgeneralization to deny that many great 30 to 40-something archaeologists have made significant contributions over the past two decades; that is undeniable.  The vague point I am attempting is that the field itself has been so fragment and impartial (dare I say aloof), that many previous quantitative advancements have not been built upon and risen to their potential.  The reasons, real or imaginary, for this circumstance are many.

A striking case in point lies directly in front of us in the list of authors autopsying archeology’s grand challenges in the pages of American Antiquity.  Kintigh PhD class of 82′, Beaudry PhD class of 80′, Drennan PhD class of 75′, Limp, Maschner, Wright are the scholars of my formative years, Altschul founded SRI in 83′, and Kohler was developing archaeological ABM when I was pushing turtles in LOGO… [right 90]. These names are the pantheon of our intellectual predecessors.  If I may be so bold, they could be enjoying their achievements and not allotting their publications to worrying about the flames that others should be tending to. Given that the challenges in the article cover a wide variety of topics, many are justified and resolved on quantitative grounds and these grounds are fertile.

** I’ll spare the armchair quarterbacking on why we dropped the ball, but this article states my opinion much better than I can.

I know I am being a bit dramatic in making my points, as there are a number of bright points producing cohesive model based archaeological literature.  First in mind is the evolutionary and biological anthro/archaeo enclaves, the best use of Bayesian methods is in bettering our understanding of temporal models, digital archaeology/humanities make use of topic models and clustering often, and Agent Based Modeling has been a great source of complexity literature.  Specialization of sub-fields is great and necessary, but if a wider audience could take the logic of an seemingly unrelated models, appraise that model in terms of the problem it addresses, and apply the same model to their own data we can begin to build a unified framework of cross-discipline knowledge.  Knowledge that crosses scales, time periods, and cultures.

Solutions? learning to code; sharing data; sharing findings; publishing negative results; identify the unique nature of your archaeological data; dedicated open-access journals; introduction to logic and quantitative methods in universities; talking to people outside of our bubbles; studying other fields/sciences; reading old literature; never being satisfied; being more skeptical than the skeptics; being creative; and adapting {stealing} ideas from the smartest people in the room…