 |
 |
Getting on another soapbox of mine, I think the lack of a general platform and standardized software that embeds key models for particular application areas is holding our community back. This is has been my motivation for the mr potato head modelling push. However a standardized model in any area needs to be a community effort that embeds the models of many research groups as special cases. Otherwise no one research group will use it, they will keep using and recreating their own models. You need a new standard model to embed a critical mass of other models.
In my twenty years in this field I have seen group after group aspire to create a common platform, working alone, and all have failed to be adopted.
We are trying to create a common model for agent based models of urban land markets. There was a lot of discussion at last falls social simulation meeting about the utility of common models and seemingly a clear indication that there are enough models of social imitation and diffusion to create a meta model in that genre.
Clearly the time is right to undertake this sort of effort for infectious disease modelling and I suspect funding would be available. And the community seems to be mobilized. So who will be the fearless leader who will coordinate the effort and make sure a common model embeds the models of other key research groups?
In general I suggest we focus on areas where we have success and what we can do constructively to improve on weak areas, rather than declaring ourselves failures.
Dawn
Sent from my iPad
On Mar 21, 2020, at 6:54 AM, Frank Dignum <[log in to unmask]> wrote:
Hi all,
I agree that we actually would need more generalized models that can be instantiated.
The main problem is that there is little theory to build those models on and people do not agree on the theory that exists as well.
I started late last year with my group to think about such a more general platform that can be instantiated. It will take several years at least, because it has to be carefully composed and argued and also efficiently implemented in a way that it can be easily instantiated.
We have taken the plunge and will force ahead. I will make no claim about whether the result will be a good usable product. But I do think that we already learn a lot by trying. Then other people might get better ideas based on this and we can generate some more quickly usable tools for policy makers or other people outside the academic community.
Note: our aim for the platform is to provide a play ground for stakeholders with which they can try out different interventions and get insights in possible consequences. No predictions!
Cheers,
Frank.
On 2020-03-20 20:02, Robert Zupko wrote:
One thing to point out here is that ABMs modeling diseases are used to influence policy (e.g., https://www.sciencedirect.com/science/article/pii/S2214109X1500162X); however, the time horizons tend to be years in the future and the questions tend to be along the lines of "How long will a treatment be effective before it needs to be replaced?" These models typically take months to years to develop, calibrate, and run to generate sufficient data to be useful though.
I suspect the biggest problem we are encountering with COVID-19 is that most of the models out there are not generalized "platforms" that we can plug an illness into, but very specialized models of a particular, well understood, illness. It's hard to develop anything for COVID-19 when the literature still doesn't have a consensus on what the mortality rate is for example.
I think everyone agrees that forecasting is not a useful activity with our models. Some contributions to this thread have interpreted the question to mean: what data and level of disaggregation should we use? Others have suggested or endorsed particular modelling approaches. Some both. One of us has suggested it is too late for us to develop anything new and useful.
I suspect that last statement is close to the truth. I wrote and published my first agent-based social simulation model in the mid-1980s and I was by no means the only one — even discounting the system dynamics literature. So why, well into or past the fourth decade of social simulation modelling do we not have the tools and expertise and common ground to make a useful and clear contribution to the most important policy issue since the Second World War and, wars aside, perhaps ever? (Think black death)
Is it possible for us all to stop coming at this problem from our own interests and the models and modelling frameworks and platforms we have developed and used? Can we step back and collectively consider why we are not able to provide clear and useful guidance and assistance to policy-makers in the course of this global emergency?
I have a proposal.
Please could we establish a group of more and less experienced modellers to produce an assessment of the reasons we have made no real contributions to policy formation during the COVID-19 emergency and produce some proposals for a development path to useful, flexible, policy-supporting social simulation modelling. Participants would need only to come with open minds and, in the first instance, to explore the reasons why we failed to make our mark on the development of policies for mitigating or containing the pandemic.
Obviously, if you think we are doing just fine or that your own next model or platform or whatever will solve all of the problems, then you would not want to be involved in such a group.
Scott Moss
Brookfold
The Wash
Chapel en le Frith
High Peak
SK23 0QW
United Kingdom
tel: +44 (0)1663 750913
mobile: +44 (0)776 968 9991
www:scott.brookfold.uk
To unsubscribe from the SIMSOC list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SIMSOC&A=1
To unsubscribe from the SIMSOC list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SIMSOC&A=1
-- Med vänlig hälsning/Best regards, Frank Dignum * Professor Socially Aware AI * Department of Computer Science * Umeå University * Sverige * Knowledge is only one point, e-mail: [log in to unmask] * the ignorant have multiplied it telephone: +46-90-7869101 *
To unsubscribe from the SIMSOC list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SIMSOC&A=1
To unsubscribe from the SIMSOC list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SIMSOC&A=1
To unsubscribe from the SIMSOC list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=SIMSOC&A=1