Battle Odds request

I've been playing about with on-line odds calculators, but they don't tend to do Barbarians. Barbarians have reduced odds don't they? Unless you're on a higher level or something?

Anyway, my question, if anyone knows for sure:

What are the odds of my Regular (3 hp) Greek Hoplite attacking a Barbarian Horseman (2 hp) that is standing on but not fortified on a Hill square (no river separation) and:

1. Dying without inflicting any damage
2. Dying but inflicting one damage
3. Winning and taking no damage
4. Winning but taking one hp damage
5. Winning but taking two hp damage

?

And:

Are potential battle outcomes calculated before the encounter takes place or during the engagement in real time?

Thanks.

Edit: Regent, Normal AI Difficulty, Raging Barbarians

http://www.civfanatics.com/civ3/infocenter/#levels - at Regent you have a 100% bonus against barbs.

The rest is from faulty memory off the top of my head:

Combat is evaluated per-round. Each round one side or the other loses a HP, so while the odds per round don't change based on HP, the side with more HP can absorb a "bad roll" or two, or in a bad match has a better chance of hitting a lucky roll or two. In other words, the per-round odds don't ever change based on HP (in Civ3), but your high-HP troops will be more successful overall.

On a hill unfortified the barb gets a 50% bonus. The Hoplite has an attack value of 1, so if you have a chance to pick your battle it's much better to have the barb attack your 3 defense value.

But you don't always have that chance, and sometimes Hoplites have to risk going on offense. Such a waste.

This is probably wrong, but I think the per-round odds are somewhat like this: The barb defends at strength 1.5 (50% hill defense bonus) and you attack at strength 3 (200% difficulty-level barb combat bonus). You have two win two rounds before losing 3. He has to win 3 rounds before losing 2; that sounds like another 50% advantage, but I don't recall the math being that simple in multi-draw statistics.

Attacking across a river affects odds, as do fortresses and barricades. And radio towers, but I doubt the barbs have radio towers nearby.

Also, I seem to recall that there maybe rounding off in some of the calculations; in this fight any round-off significantly alters the real odds. Long ago people did lots of trials and determined these things and posted their findings. It's probably still around here somewhere.

Thanks for the excellent response.

So, no matter how you cut it, the Hoplite should still have the best odds in such a scenario, even with all the rounding involved?

Yes, it has the upper-hand, I'm guessing roughly 2-1 or slightly better odds, but even that means you lose 1 out of 3 or so, but I always feel several times more angry about losing than I feel happier about winning, so the overall pay-off odds are negative.

I was being silly about the "pay-off" odds, but now that I said it, it leads to a more important train of thought. For the sake of argument let's say the Hoplite wins 2 of 3 of these battles and loses 1 of 3. You spent 20s on the Hoplite, and it's 3x as effective on defense. If this were an enemy civ and you repeatedly fight this battle (an enemy civ that keeps sending 2HP warriors your way? eh, work with me here), you both come out even on average on resource expenditure. You each lose 20s worth of production every 3 battles, on average. However, barbs spawn and you never diminish the "barb civ" by killing their warriors 2 out of 3 battles, and there is no payoff except minor ones like promotions. The only reason to risk non-optimal battles is to protect your civ's other assets.

You probably know what the barb is immediately after and can probably tilt the odds more in your favor by positioning for a more favorable battle by either having the barb attack your hoplite or defending the worker or city with the Hoplite.

However, if the barb is in a position to immediately take out a worker, settler or clear a lot of shields from a city build-box, you might rather take the chance of losing a Hoplite than losing the production.

Oh, totally, I never normally use Hoplites for Barbarian bashing. I normally do the wandering Warriors routine and then send either Spear/Arch combos or Horses for the latter Barbarian stages when they start to pile out their camps.

However, I was trying something new today, just mucking around really with random experimentation and I was doing a 'just Hoplites' experiment (that wasn't the only experiment but that's the only one that matters for this thread), so I was getting in the habit of saving before doing something random.

I had cleared out every Barbarian in my known area (about 15 of them, 12 of which were just Barbs attacking, and losing, to one Hoplite, lol) apart from one Horse that managed to end it's turn in an awkward position where it felt the right thing to do to finish it off with a (sole) city-stationed Hoplite - because that was the only unit that could reach the Barbarian that turn - the horse could move/pillage next turn if I just left it until next turn.

So I saved and then attacked the Horse with my Hoplite. I lost outright all 3 hp bars and died with no loss to the Horse. Typical luck, lol, I thought and reloaded. Same again. Reloaded, same again, same again, same again, same again, same again, same again - as if the battle was either pre-destined or somehow impossible odds.

So I'm thinking it's either a bug... or... what exactly? I exited the game and reloaded the game, same again, same again, same again.

Anyway, I kept that as my save point and, if this happens then next time I fire up the game from a position of a fully shut-down computer, I'll post the save and see if someone else can get a more varied set of results from the situation. If it ceases I've at least narrowed it down to something solvable, but if it continues I'll post more.

Ah, you are experiencing "preserve random seed". If you save, do a sequence of events, then reload and repeat the same sequence of events the "random" numbers come out the same every time.

If you're reloading for a different result you have to change the sequence of random events. Usually in a war you reload and attack in a different order. In your test I can't think offhand of an easy way to spark a random event to change the combat odds for that fight. Try stealing a tech maybe? Bombard a square with any artillery unit? And then attack again, and the results may well be different.

Most computer "random" numbers are really pseudo-random and will repeat results when starting with the same seed. Somewhere in the settings is a "preserve random seed" you can turn off, too.

Ahhhhh, that explains it. I always wondered what Preserve Random Seed meant, I always thought it just meant saving the Map Seed for the Play Last World function, I had no idea it was an anti-random in-game device.

So is it a tool to prevent save-scumming? Or is it a tool to be of assistance in the advent of a Crash To Desktop?

The phrase Preserve Random Seed in the context of how it works seems like a complete contradiction in terms, an impossible sentence - non-random random.

I think it's primarily anti-cheat. If you can reload every time you lose a battle and try again until you have better results.... But of course you can turn it off--it's an option during the start new game screen--and you can alter history by changing the sequence of random events on reload.

Random number generation is kind of interesting. The trick is in getting enough entropy to be random. The easier way is to get enough entropy for one random number and seed a pseudo-random-number generator which have predictable results if you know the seed and algorithm, but in a game situation they work out just as well as random numbers and have the advantage of being able to repeat pseudorandom sequences which make it possible to recreate the same map with a seed given the same map generation options, for example. And in Civ it is used as a cheat deterrent.

So if you've ever done any programming and used random numbers at all, "preserve random seed" makes sense. But it's not intuitive to non-programmers, I guess.

Interesting stuff. I've since pressed end-turn and the Horse goes away and attacks a different Hoplite, so in this instance it changed nothing about the game. Still, it's curious that they felt they needed to enact an anti-cheat into a game like Civ, because Civ already has soooo many player exploits and cheats that isolated battle outcomes are really the bottom of cheat food-chain for any player. In any one game that's taken to it's final conclusion there's probably maybe one or two encounters, out of thousands, that would tempt someone to repeat individual fights.