Damnit! Beat to the punch by the 15 year-old yet again! Using the physics engine to determine things like the rate of spread of fire is overkill. It would make more sense if the world of Elemental were continuous and not grid-based, but it is grid-based and thus determining how things spread doesn't need to be that complex. To determine how fast fire could spread, they could just have some parameter for rate of spread (# of tiles/turn, modified by a little bit of randomness and by type of terrain - it should also be probabilistic imo). This would be just as effective as using the physics engine, but much easier to code, much easier to mod, and much less computationally expensive. 

I don't understand what you're saying. Could you explain what you mean regarding the grids in clearer words?

And, I don't think it's as simple as saying "let havok do it!" For one, as far as I know, havoc doesn't simulate fire (maybe I'm wrong?). Secondly, having a physics engine that determines things like the rate of spread of fires IMO is overkill, unless you try to make it actually realistic - in which case the simulation is going to bog down slower computers. Secondly, I'm not sure how a physics engine would be used to simulate something in turn-based time. Physics doesn't work that way - the only thing I can think of is to have each turn last a certain number of physics engine seconds, but that could cause huge problems (different processes taking place physically on totally different time scales, for example, among other more subtle issues).

Having a physics engine deal with dropping a mound of dirt from the sky and turning it into a hill, spreading fire, and all those things over the map seems like it'd both be overkill and would in some cases detract from the game. The mound of dirt in particular, as in the end it wouldn't look like a hill it would look like a mound of dirt. Having physics tied into spells and such on the tactical map would be very neat, though. I'd love to to see fireballs, lighting and explosions throwing units around.

I see no difference. Whether it's a regular grid, a grid-within-a-grid (which is essentially just a smaller grid!), or some small variation on a theme I think just having things like rate of spread of fire be determined by a spread factor modified by some randomness, terrain type and other enchantments will work as well as using physics but with much less overhead. Such a method will look more than good enough, function wonderfully and be very simple to calculate and display.

Game physics is just calculations, because physics itself is just calculations Even if havok is capble of dealing with fire, you are completely right that it would require the source, initial fuel, potential fuel in the vicinity, wind direction... It'd also have to know distance between fuel sources, factors that would put out fire, etc. For one, this would require every object or at least every terrain tile to have some sort of 'fuel' property, it would require them to put in Wind direction for every tile - having factors like these (and many many others if they want their spell system to be very general) would be insane (and having this many more parameters to track for each and every tile and possibly object would probably not be appreciated on computers with less memory). AND there's the whole part about, being tile-based, each discrete portion of land must either be on fire or not on fire, and it would evolve in large discrete intervals: using physics for this would be a crazy amount of overheard for no gain over a simpler system...

You could easily adapt a physics engine to work on a turn-based basis, but there isn't much of a point. You won't even notice the physics happening if it happens in turns. And even if SD could manage to solve the time-scale problem (which I suspect wouldn't be that easy to do), there's also the issue of the fact that discrete physical evolution using large time intervals can result in undesirable effects - important or expected effects but not ever occur simply because they happened somewhere inside those long intervals. A fire that spreads or burns fast enough could result in a tile being pristine one turn and burnt to the ground the next, without ever having been on fire, for example.

That is not completely true. If you want to drop a pile of dirt and have it dynamically turn into a hill, shaped and sized based on the amount of dirt and the elevation and shape of the land below it, then the dirt must be treated as lots of particles by the physics engine (otherwise it just wouldn't work right). After stressing your computer for a bit, the dirt would fall and make a convincing mound of particles. The graphics engine would then have to render all of this - but during the falling process it'd have to render the individual particles, but once they settle it'd have to somehow make them look incorporated into the terrain. You lose some graphical freedom for any animation that uses a physics engine, particularly if you want something to be fine particles one moment and a cohesive solid mass the next (the other way around isn't nearly so difficult, though).

That is not true. Having physics determine the spreading or burning rate of fire affects gameplay. Having physics determine how large a hill and what shape your magically dropped ball of dirt will become affects gameplay.

Using Havok in as many places as applicable increases the amount of worth you get out of it, but using it in places that you shouldn't just because it'll increase the worth you get out of it is downright silly. Game physics is a wonderful innovation and can be used to better gameplay and graphics to enormous extents in many situations, but there are still many situations where physics will not help, or will not help enough to make the extra computing power or memory requires worthwhile. The beauty of physics is in the exquisitely fine details. Physics is wonderful for smashing walls at 60 fps, where a non-physics method would look unnatural. However, if you smash that same wall at 2 or 3 fps, the differences between each discrete state are so large that you aren't gonna notice the difference. That is why things like turn-based fire spreading will gain nothing by having it use physics. Physics will allow a huge number of dirt particles falling from the sky to form a realistically shaped hill based on terrain factors (though it'd take some time to calculate if you include enough particles), but it would be supremely difficult for the graphical engine to render that information in a pleasing way (it could display the falling dirt well, and the final result well, but animating the transition would be hairy at best) - and I think a non-physics approximation would be more than sufficient anyway.

Well, with the fire example: there is no wind (because i don't think that sort of weather detail COULD be implimented), and everything is limited to binary squares (ON fire or NOT ON fire), so the physics engine would have to perform the necessary calculations, then "round down" the results to the ON or NOT ON state for each square. And the result would be exactly the same: after a short, quasi-random number of turns, each flammable square that borders the flaming square would catch on fire. But it would be faster and easier on the dino computers to just have the code say "after a short, quasi-random number of turns, each flammable square that borders the flaming square will catch on fire", than going to all that trouble.

Setting fire to a forest does one of two things from my point of view it either destroys a resource or damages building / units inside if forests are passible terrain or if they are impassible it creates a new path to march troops or whatever. The physics calculations for this have no impact on game-play. The hill / mountain creation is similar it creates a natural obstruction to the player that either needs to be removed or gone around. The physics calculations have no impact on game-play. I hadn't assumed you would be able to drop the hill / mountain onto towns, create an earthquake, or something of that nature. 

You misunderstood my statement which is partly my fault I should have worded it as follows. "The respective calculations for fire propagation and hill / mountain creation would largely be the same with only a decent amount of unique calculations for each action." As for extra properties being added once again I feel you are making it more complex than necessary. You wouldn't need that many variables / properties for each tile. Using Havok that option would always be there if you wanted to do more advanced things in the future but just to get the basic functionality we are talking about it would not be required in any shape or form. 

As for the video I looked at the guy's specs we are the same in every area except his processor is better (I have an E6400) and I have 4 GB of memory. My gaming computer is considered by the vast majority of gamers to be average / below average. As I said before I disagree with your point of view on this subject and there really isn't much more to be said. If I find some more data relevant to this conversation i'll be sure to bring it up. 

Lastly, mathematics is "one" of the ways we use to understand the universe. I don't personally consider it the only or best way. 

The distinction is: while tiles do REPRESENT continuous terrein, they are not, in fact, the same thing.  A tile exists in any number of "states": on fire, covered in grass, has unit Y on it, etc. These states can overlap, but there is little in-between: you can't have a "knight on bottom-right part, upper section is on fire, lower-right corner has a tree" state. Thaus, it is simpler to just do things the old fasioned way tile-by-tile.

I just don't see how you think it will affect game-play. If I have a 10 tile map 5 of which have forest on them and I set one forest tile on fire. Using physics based calculation the fire would spread to other tiles adjacent to the one that has been set on fire. The path fire takes is very predictable and linear in nature. It couldn't jump over a tile unless a programmer took the time to create such an effect. Even if you expanded this to 1000 forest tiles and set the center one on fire anyone could accurately predict how the fire would move through the forest. It would expand in a circular shape until it consumed the forest. For the purposes of a game if there are no advanced weather features you could even ignore wind direction.

You could make a system to simulate this effect not using physics which would produce largely the same results with irrelevant changes in process.  I don’t think the physics engine approach of choosing which tiles to set on fire based on a simply calculation vs. a randomized system would effect game play. The end result would be the same if both were set to have the same constraints. If a LV 1 fire ball spell has the constraint of only burning a maximum of 10 tiles the process for each system would be different but the result would be the same. Hence it would not affect game-play.

Hill / mountain creation would have the same result it would not matter if you used a physics calculation to create it or a static system. If LV1 hill creation could only create a hill that was one tile big once again the processes for each would be different with the end result being the same. The only difference is with a physics calculation system you would get unique looking hills while with the other you would only get a number of pre-designed hills. This would not affect game-play.

I think you’ve overlooked the idea of constraints on everything. As long as the constraints are the same for both systems it doesn’t matter how each does it as long as they do it within the constraints. The resulting systems would only impact game-play if they had different constraints which I never suggested.

On the subject of physics I concede mathematics is the only method used by some people. Personally when taking physics classes looking at an equation or the math behind something rarely helped me understand the nature of an event. I always needed to understand potential reasons why first then the math would fall into place. I also disagree with physics being completely limited to mathematics. 

Ok let me make myself perfectly clear. I will state the advantages as I see it for having a physics based system for such effects.

1. Negligible increase in calculation time if done properly.

2. The code is already there to be manipulated thus saving time.

3. Allows for potential expansion to more advanced effects in the future. 

4. You paid for it might as well use it.

5. I have Intel stock, Intel owns Havok, what is good for Havok is good for Intel, What is good for Intel is good for my portfolio. (joke??)

Agree / disagree / whatever I've enjoyed the conversation none the less. 

4. Huh? I don't quite understand what you mean?

I was making reference to the fact that Stardock had to purchase Havok's physic engine. Since they paid for it and it would be well suited to this situation in my opinion they might as well use it.  

Edit: @ Scoutdog I would replace the word fact in your post with potential. 

It appears this is largely a misunderstanding of what we mean by "gameplay." What I mean is, unless you intend to make an extraordinarily simple method of propagation - like fire spreads 1 tile per turn, always - the physical simulation and the 'regular' methods would have slightly different results. Even with the same seed, one tile that burns in one method might not burn in the other, even if they achieve very similar end results. Two people playing duplicate games, doing exactly the same thing as each other but one using physics for these things and the other using the regular method could end up playing somewhat different games. Small differences propagate into much larger ones - it's why desyncing is such a problem: a small desync doesn't stay small, it hurtles out of control.

And regarding hill/mountain creation - your example only works if such spells are constrained to affect one tile. That would suck horrendously. And without such a constraint, the results of this particular spell would differ wildly depending on whether or not physics is used. Secondly, without a physics calculation system you are not constrainted to a number of pre-designed hills. Do you think the hills and mountains generated by the random map generator will all be predesigned? It doesn't look like it from the screenshots we've seen so far. The exact same algorithms could be used.

The means to the final result are as important as the result itself. If one method burns down a forest faster, or in a somewhat different order than the other, then game-play is effected, even if only subtly.

Just because you needed somebody to explain the equations and mathematics to you in a language you are more familiar and comfortable with does not mean those explanations are fundamental. Especially when you get to the higher levels of physics, most people find that they start looking at the equations to clarify people's spoken explanations more often than the other way around. Also consider that in very close to 100% of situations, the mathematics precedes the conceptual explanation (which is really just an interpretation of the math into another language). Please, don't have this argument with a physicist.

Regarding your later post, Scoutdog already covered it.

Your words:

And where did you get this understanding from? Either someone explained it to you, or you discerned it from the mathematics. Trust me - if you can't put an idea into math, it ain't physics 99% of the time. There are very few exceptions - and most of them that I can think of off the top of my head are in fields like astrophysics. But even there, observations are ultimately turned into mathematics - it's just a rougher process that usually takes longer due to the nature of the beast. You can believe as ardently as you want that this isn't the case, but you'll be wrong.

Now, when it comes to the game, your opinions are at least as valid as mine And considering we've already pretty much gone in circles I think you're right that we should leave it at that - at least until we know more.

Um.. the 'logical theory' behind numerical calculations IS math. Yes, anybody can plug numbers into ∮E•dA = Q_enc/ε0. The logic behind that calculation, however, is in that equation I just wrote. Even if you've never seen this equation before, if you know what is represented by the variables an astute or experienced mind can figure out from the math itself that it means that the electric flux through a closed surface is proportional to the enclosed electric charge. Or, if I generalized the equation to ∭(∇•F)dV = ∯F•ndS and providing the meaning of the variables, one could deduce from the math alone that it means that the net flow out of a region is the sum of all sources minus the sum of all sinks. All of that information is encoded into the mathematics. Theory is where the real heart of physics is - but physics theory is mathematical in nearly every single case. The only aspect of physics that is not mathematized is observational physics, where the conclusions drawn are along the lines of "this is what happens when you do this" - but even there the goal of that observation is to figure out a mathematical model that explains that process.

Which is one of the reasons why we don't have a clue what the heck happens inside a singularity. Conceptualizations of what might occur inside a singularity are not really physics - they're conjectures made for the sake of curiosity without any real evidence or chance of validation. Until someone comes up with a mathematical framework that does not break down inside singularities or some other workaround (or, even better, a physical model that does not degenerate into a singularity in those circumstances!), physics will continue to be inapplicable to singularities.

The conclusions in physics are the equations themselves, and the logic behind them is the relevant data and preceding theories (equations!) used to derive them. The logic behind numerical calculations are simply the equations used to make said calculations.

Sorry for the off-topic...