Dispersion sucks

[fran]

The map generator is unfair enough by creating different land and settlement locations.
Dispersion=6 as is now means that one player might have more than one settler on one tile, all tiles with settlers close together. Another player might have all settlers scattered far apart. The last variant is clearly a big advantage if you happen to find yourself beween mountains, hills and forest. All unequality summed up you can be really screwed and somebody else can be quite happy.

Therefor I suggest to disable dispersion, meaning all units start on same tile. That's probably dispersion=0.
I understand that doesn't compensate for being between mountains, hills and forest while somebody else
is on long river with plains, wheat and oxen. While dispersion doesn't solve that unequality it could make it worse.
Of course it also could mitigate it slightly. But in my opinion that's to much chance for the beginning where we have to much chance anyway.

[Corbeau]

Actually, it's all a matter of probability and smaller dispersion means bigger deviation from the average. With all units on the same tile, it is absolutely crucial how that one tile is placed. With units scattered around, it is much less probable that ALL will be in a shitty position and it is also much less probable that all will be in a great position.

[fran]

Actually, it's all a matter of probability and smaller dispersion means bigger deviation from the average. With all units on the same tile, it is absolutely crucial how that one tile is placed. With units scattered around, it is much less probable that ALL will be in a shitty position and it is also much less probable that all will be in a great position.

That's the usual answer to my proposition, and I didn't expect sth different, though I still think your argument doesn't hold, because the map generator creates clusters of certain types of terrain that are usually larger than
dispersion=6.

And I'm not against dispersion in general, but against the current implementation. Dispersion could be the same
for everybody, but as I explained in my original post the current implementation will randomly set the units, which results in good dispersion for one player and bad dispersion for another. There is no need to introduce this additional chance that -- for some players -- cumulates with all the other chances to very good luck vs very bad luck.

My starting point is that the /starting position/ should offer the same chances in the game for all players, as much as possible. The more chances you stack onto each other by creating the starting position the more you increase the absolute spread between the luckiest and the unluckiest player.
For the sake of simplicity let us assume that every chance will give you "bad" or "good" (it's not like that, but that doesn't matter for the argument). So with one chance half of the players will have bad and the other half will have good. Now introduce a second chance, that gives 4 results: good/good, good/bad, bad/good, bad/bad. That's fair for 50% of the players (namely good/bad and bad/good), but not for the other 50% (namely good/good and bad/bad). Even if you introduce a 3rd or 4th chance, there will be players that have good/good/good/good or bad/bad/bad/bad. I don't understand why one wants to do that while creating /starting positions/.

Be that as it may.

What about setting the IgTer flag for settlers? That would ensure that terrain doesn't matter.
(Except rivers, that is. But it would mitigate the problem.)
The settler unit could be cloned to one that has IgTer added, but this type of settlers could not
produced in-game but only reached out as start units.
With the IgTer flag set dispersion could be switched off and it should be as fair as it gets with the
current implementation.

I started LT40 with 5 settlers on 3 tiles, all close together. No big chance to evade forest, hills, mountains.
That means your settler will move 1 tile per turn. You know how long it takes to spread them?
My neighbour travelled gently on long long rivers.

[wieder]

Not a bad idea to set the igterrain for the settlers. I can't remember what was the reason why it wasn't done before? Or was there one.

If I remember correctly dispersion 0 might lead to situations where the players starts on a very narrow or small starting position. Might have been with 2.3. Not sure.

[fran]

If I remember correctly dispersion 0 might lead to situations where the players starts on a very narrow or small starting position.

That's actually a good point. If the placement/dispersion algorithm doesn't check for isolated/coastal/edge positions a higher dispersion is better.

[Corbeau]

Actually, good points.

AND:

What about setting the IgTer flag for settlers? That would ensure that terrain doesn't matter.
(Except rivers, that is. But it would mitigate the problem.)
The settler unit could be cloned to one that has IgTer added, but this type of settlers could not
produced in-game but only reached out as start units.

this is exactly the idea I had for my ruleset for the Founders unit. The problem is that you can't have a "can't be produced" unit as a starting unit. Literally. I put all the necessary flags and the game simply wouldn't give me Founders. instead, I got regular settlers.

I don't remember how I found the explanation, either somebody told me (although I can't remember who and where, I searched the forum.freeciv.org and can't find it) or I found it in some of the manuals somewhere.

Anyway, the idea was first mentioned here:
http://forum.freeciv.org/f/viewtopic.php?p=80766#p80766

I elaborated it a bit more in the thread with my ruleset but eventually, as I realised it can't be done, deleted it.

Another variant was that the Founder is THE starting unit and taht it gets obsoleted by Settlers, but then you need to choose the tech to obsolete it and then it becomes messy because some people can take other paths.