It was found that for battles <24 hours, the predicted casualties were too low. A couple of changes were tried, and if I read the article correctly, consist of the following:

1) battles less than 4 hours are treated as 4 hours
2) battles less than 24 hours have casualties = SQRT(fraction of 24 hours) as opposed to (fraction of 24 hours)
3) fanatic combat results in casualty rates 250% of normal

Reading through the article, I wasnt sure if #2 was still in the mix, or was it thrown out by the end of the article ?

thanks,

I grabbed all the newsletters and started digesting them, looking for relevant information.

Do you guys sell the different databases separate from the TNDM contracts ?

thanks,

quote:

---

Originally posted by Mad Dog:
[B]So, as long as the combat time is >4 hours, battles < 24 hours have linear fractional losses (8 hour battle = 1/3 losses) ?

---

Yea, it is an uncomfortable modeling compromise, but it seems to work. The main problem is that a lot of battles have a short intense period amid lots of less intense periods, so by running the models in really short time increments, it distorts the results.

quote:

---

I grabbed all the newsletters and started digesting them, looking for relevant information.

---

I can email you the latest two newsletters. We have not updated this site since 2002.

quote:

---

Do you guys sell the different databases separate from the TNDM contracts ?

---

Yes, but at prices for large organizations. It is out of reach for individuals. Some of these databases have hundreds of man-hours of labor invested in them. For example, our division-level data base has more than 740 cases in it now. If one figures that it takes and average of 3 days to create a case, then were are looking at something like 9 man-years invested in creating that data base.

thanks,

Mad Dog