Saturday, 29 November 2014

Top Down and Bottom Up

There has been some discussion recently about the top down and bottom up views of rule writing, and the implication of this for the design and functioning of the rules themselves. I suspect (and will attempt to show) that this is a cultural thing, or possibly just another case of the ‘scientific worldview’ invading other areas.

So far as I recall, the bottom up view of rules was the first out of the blocks. Certainly, I remember reading, I think in Charles Grant’s ‘The Wargame’ comments about speed of movement. These were along the lines of a normal person can walk at about four miles an hour on a flat, level, hard path. Thus a unit can move however many inches determined by the ground scale and time scale of the moves. This, again as I recall, is arbitrarily reduced to something reasonable, on the grounds of having to keep in line, the various bits of a unit having to march around blackberry bushes and so on.

So this is a bottom up design of movement rules. We take an individual and assess his capabilities. Then we work out what implications there might be of loading him up with eighty pounds or so of equipment and expecting him to keep in line with his fellows. Even then, we obtain a number which might be a long way from anything with is either reasonable in wargame table move distances, or historically verifiable. Units moved slowly, relative to an individual.

So, of course, Grant (and he was not alone; he is just the author I remember) was well aware of this and fudged the results. In the end, then, it could be argued that despite his explicit bottom up approach to rule writing, he resorted to a top down approach to get the ‘right’ answer.

Such trajectories could be multiplied. For example, there were results of tests around (from, I think ‘Firepower’) of shooting muskets at unit sized sheets, and working out the effectiveness. Despite the author’s charges to the effect that this was an absolute, theoretical maximum, I fear some wargame rules dived in with the idea that muskets were something like 60 – 80% effective, and the body count in wargames rose as a result. Even though the effect of being in battle, being under fire, accidentally shooting out your ramrod and so on were commented upon, rule writers took the headline figure and worked with it.

Now, the people who did these things were not stupid. The thing is that they look as if they are scientific. We all like numbers; they give such an air of authority. A recent incident at work indicated this rather well. A student had done a very nice project and got some clear qualitative data. This was insufficient for her supervisor (who really should have known better) and she was told to do statistical analysis to prove the point. The problem is that statistical analysis on qualitative data is, well, asking to have the numbers made up for you. But numbers have authority where words do not.

Part of the reason for this is, I think, the success of science. Science give numbers; when I was a student you would derive a formula and then ‘plug the numbers in’ to get the required answer. Of course, what everyone knew and tacitly ignored was that the numbers were made up to make the answer pretty. As with so much, even at an undergraduate level, the experiments and problems are fixed so that they ‘work’. That, after all, is how science is supposed to be.

Real life, however, is messy. As an experimental physics researcher, plugging the numbers in became a game a bit like accountancy. If you ask an accountant what the values of a fund is, one of a set of possible answers is ‘what would you like it to be?’ So it is with experimental physics. The question is not ‘what is the answer?’ but more along the lines of ‘how can we get an answer, and how reliable might it be when we have it?’ Numbers, even the outcomes of equations, give us spurious confidence. As someone told me once, ‘it isn’t the answer that is interesting, it is the error’.

So, in wargame terms, we are probably better in going top down, in looking for how a body of men actually performed under battle conditions, be that in movement, shooting or whatever else. This, too, has its dangers, of course. Firstly, the evidence is, to put it politely, patchy. Mostly it does not exist. Where it does exist, we are probably back to those parade ground performance figures which are a guide to a perplexed young officer, but not much use to the old hand sergeant. He knows from experience, and it is not written down in a book.

The problem is, then, that we can neither be purely top down nor purely bottom up. Our records of unit performance are based on individual observations, and may not be valid for all units, let alone all times and space. Top down views are therefore contaminated by, at least, particularity. This unit did this in this battle, so we universalise an individual performance. Of course, bottom up is no better. It takes no account (except through fudge factors, as already noted) of emergent behaviour and bears even less relationship to real life than a top down approach.

But perhaps the major difficulty with the top down approach is the fact that such views of the world are frowned upon, culturally. Science, or at least the perceived method of science, is king. And science is irrefutable (nearly) reductionist, and hence bottom up. The ‘Great Chain of Being’ with everything in its place from worms to slaves to women to men to angels and then God himself has irretrievably broken down (I’m not saying that this is a bad thing), and the legacy of that is that everything now has to be bottom up.

It is just that for writing wargame rules, it doesn’t work very well.


  1. The origin of the bottom up approach for war gaming as far as we have evidence at this point is von Reischwitz in the 1830s, and were designed for training officers in the difficulties and confusion and command, not for entertainment. They were based on contemporary manuals and tests crossed with experience of such warfare not on patchy historical guesswork like modern hobby authors. It appears that they worked fairly well as long as you had large teams of assistents and players and experienced, capable intelligent umpires and time to work out the 100 or do turns to see the results of orders. The concept is still seen as valuable in military and business circles but the detailed processes have developed and of course computers help.

    Other than that and a few details, I don't substantially disagree though I am nit convinced that all if our actions are the result of a subconcious scientific view or where we could now find humans uninfluenced by it so ad yo have a control group to write a set from scratch.

    I do agree thst most war games are a mix of approaches but I very much disagree that they are all now more bottom up. If anything the trend to day is "design for game effect" with little or no heed to the details needed for a bottom up approach.

    1. I'm sure a lot of our actions are not subconsciously scientific, but in modern culture 'science' is the paradigm of knowledge. Thus getting people to answer questionnaires and giving the answers as a percentage passes, in popular culture as science, knowledge and / or fact.

      Thus, recently, it was reported that 19% of respondents believed life came from organic molecules from comets, and 15% believed it came from God. 'Science outstrips god' was the headline. what about everyone else? And are the two questions actually incompatible anyway. But stick a '%' afterwards and it looks serious, official and factual.

      I do think that most recent rule sets are top down, incidentally (DBA etc). But the problem is that someone still has to move the units. perhaps some micro-motors could help me here?

  2. This (great) entry reminds me yet again of the inaccuracy of shooting - but also the phenomenon of the 'baseline'. We absolutely know from a 'top-down' approach that it takes loads of arrows/bullets/shells to cause a significant number of casualties, but we know even more reliably that a well-trained archer/musketeer/rifleman can hit the target almost every single time at 25m for the older stuff, 100m for the modern stuff. And once that gets in your head, even when you fudge the figure downwards, it is hard to get that initial figure - the baseline - to disappear from your calculations entirely.

    And although Kriegspiel does seem to work fairly well, I'd argue that its casualties are pretty high (military wargames and exercises have a hard time getting away from that firing range/manufacturer's specification too).

    1. I must try to did out Bert Hall's book; as I recall he has a long discussion of why musketry is so inaccurate. Something to do with the ball bouncing off the inside of the barrel and hence emerging from the muzzle at unpredictable angles. I guess it is different with modern weapons, but muskets are inherently inaccurate.

      Most wargames seem to have casualties too high. I'm not sure why, but say, DBA has (I think) a 1/3 lost rule. Now granted, not all those units are dead or wounded, but most armies still had more than 2/3 remaining effectives when they lost...

    2. I seem to remember one of the first sets I ever played with back in the 70s at our school war-games club. ACW infantry fire was absolutely lethal and opposing lines never seemed to get closer than 300 yards because below that they just got shredded. Clearly a bottom up set and clearly pre-Paddy Griffiths.

    3. Ah, the massacre school of wargaming...

      I suppose that ACW musketry is rated as better than, say, Napoleonic. so if you have overestimated the latter, the former must be getting towards a machine gun...

  3. Muskets are inherently inaccurate, but in testing they were still able to achieve creditable results (modern weapons are usually more accurate than the person firing them, the reverse being true for smoothbore muskets). It is just that the amount of casualties caused seems to bear no relation even to this figure.

    In relation to earlier periods, the casualty figures just don't seem to suggest to me that large numbers of men were engaged in "fighting" as seen in films, so you simply can't deduce what happened from the arms and armour of the period,whatever their lethality. I guess.

  4. I find it interesting that the author concludes with a paragraph that states "Science is king". It may be true in the world of gaming.

    Outside the hobby social observers are identifying a trend that's being described as the "retreat from reason". News reporting and public discussion is either misunderstanding or ignoring facts, numbers and science.

    At the harmless end of the spectrum this means people plugging their books on daytime TV telling us why there were dinosaurs on Noah's Ark (which reminds me - I must get around to gaming that). It becomes rather more serious when politicians can reject hard numeric evidence simple stating "I don't feel that's correct".

    At its very worst, people who should know better are tempted to "Sex up" reports. Sometimes thousands die form the consequences.

    Following all that: Here's my tuppence on bottom up vs top down.

    Bottom up: Here's what a man can accomplish in 15 busy minutes.
    Multiply by 100,000 for what a Corps can accomplish every hour.

    Top Down: Here's what a Corps accomplished in 6 hours, Divide by 150 for what a battalion can accomplish in an hour.

    I'm inclined to prefer the top-down model provided sufficient data is available (for example March rates of Napoleonic Corps are well documented. The effect of doubling the cavalry screen ahead of a marching corps yields a set of distinct results, often very dependent on local conditions.