I did some further research in the matter and tested a few of the earlier cities, Nubt, Thinis, Perwadjyt, and other cities of which I had saves just before the finish. By adding a few buildings or deleting a few houses it is rather simple to vary the population size and treasury upon finish and test the result. Also difficulty can be changed rather easily and Nubt and Thinis are such short missions that they can quickly be played at full length, and investigate the effect of time. Moreover because they are short, the effect of adding a few months is relatively large, so it can be established even if there is still some small uncertainty left in the score. Changing of the other factors, Culture etc, is more difficult, and more importantly, the effect of those is very small. I have collected my data so far in an Excel file, anyone interested can send me a mail and I will send the file. The results are not yet conclusive, this is what I found so far
The hypothesis was an equation of the form
Score = ( Wm* Money + Ws* Pop.Size + Wc* Culture + Wp* Prosp. + Wm* Monument + Wk* Kingdom ) * Diff / Timewith Wm..Wk fixed weight factors, a factor for difficulty and the time in months. That is definitely not the correct equation for Cleo. The difficulty is indeed incorporated by a factor, for VE Diff=1, for E Diff=2 etcetera (Note that I have corrected the above post). It also seems that the score is inversely proportional with the time in months, I have tested Nubt for 4-8 months and Thinis for 9-14 months and the results appear to corroborate the assumption.
To investigate the weight factor for money, I have placed some extra statues just before the finish, and made plots in Excel of Money*Diff/Time against score. The result is a straight line with slope Wm, and intercept the effect of the other parameters, which are kept constant. For all cities within round-off error I found Wm=10. Of course the intercept may include the effect of starting funds, which obviously are constant too. However, taking the backup loan has no effect on the score, i.e. there is no penalty, the money is taken into account straightforwardly. If you want to increase your score, it is often better to go broke in a short mission. The weight factor for population size is investigated similarly. A difficulty is that changes in the population size often lead to (small) changes in the treasury. Hence I have plotted Pop*Diff/Time against Score-10*Money*Diff/Time, the score corrected for changes in Money. Within the range of population sizes investigated I do indeed find linear fits with slope Ws, but it appears Ws is different for each city investigated. For Nubt Ws=0.145, for Thinis Ws=0.45 and for Behdet Ws=8.2. So if the equation is correct, the weight factors differ from mission to mission. Moreover the weigth numbers are non-integer and differences between missions are large. Of course the equation may be wrong, and the weight factor may depend on the population size itself. After all the population size in Nubt (200), Thinis (400) and Behdet (2500) is quite different. To further investigate whether the hypothesis of a constant Ws can be upheld, I tested Nubt for substantially larger population sizes by artificially depriving a large number of houses from food. It indeed appears that for higher population sizes the equation as given above, with the slope Ws=0.145 and intercept as determined for small population sizes underestimates the reported score. Whether the slope reaches the value determined for Thinis also for Nubt at similar population size I have not been able to establish conclusively yet, it seems to be lower still. Now the linear fit has been proven incorrect, the most obvious modification is an equation with a term Ws1* Pop + Ws2* Pop2, but if those weight factors would differ from mission to mission there is very little predictive value, if the equation has any value at all.The Money and Population are by far the largest two terms in the score for any city, so it is difficult to investigate the effect of the other terms. As far as I could tell from the data for the larger cities the weight factors were (substantially) higher than for the Pharaoh case. Also it seemed the weights Wc..Wk were not the same for all cities, or they depend on population size too.Within the range where the data fitted the linear dependence on money and population for Nubt, I checked what the effect of the only remaining parameter, Kindom, is. Throughout the tests all other parameters were zero, and Kingdom took values 40 and 50 only, so this not a very thourough investigation, but the data fitted with an equation
ScoreNubt = ( 10 * Money + 0.145 * Pop + Kingdom-40 ) * Diff / Time The weight factor Wk=1, like in Pharaoh. It seems there is an offset on the score, but with the uncertainty about Ws and the fact that there are only two independent data points (I wouldn't know how to generate more in Nubt) also this may not be the last word on this subject.On the other hand maybe it will. If the Cleo equation is indeed more complicated than the simple form found to adequately describe scores for Pharaoh, than given the low importance of the issue I doubt whether anyone will be willing to give the subject much time. I do not. The answer can simply be obtained by asking the designers, if they would remember, or by reverse engineering of the program files, if that weren't forbidden by the EULA.
The main conclusion is that the Cleo equation gives the same weight to money as the Pharaoh equation, but for the larger cities the weight of population size is increased. I have not tested Pharaoh, but if I can trust Ben the Vizier's conclusion the weight in Pharaoh is 1.5. For Cleo smaller weights are found for small cities, but substantially higher weigths are found for larger cities. Maybe the same is the case with weights for the other parameters.
[This message has been edited by joshofet (edited 06-23-2004 @ 08:37 AM).]