[NO PLAYERS] Seeding and ranking discussion

[Commodore]

Any more discussion on this? Could start spitballing locations if so...

So I think our rough consensus is as follows on players?
Will win unless we set him to Deity and the rest to Settler:
1st. OldHarry

Without Old Harry, would win unless we set him to Deity and the rest to Settler:
2nd. Gavagai

Good:
3rd. 2metra+OT4E
4th. REM
---Some debate about which is better, but OT4E is good, think this ordering is fair---

Medium:
5th? ipecac
6th? Alhazard
7th? Molach
---MUCH debate here; I'd actually say ipecac>Alhaz>Molach for sandbox skill, but Molach>Alhaz>ipecac for war tenacity---

Lead boots:
8th. Dantski
9th. Greywolf
10th. TheWannabe
---All going to be food, but edibility probably is in that order---

[Commodore]

Here's what I think the starts should be rated.

First, some notes about land quality, in addition to what what we've already said about strategic position on the map and luxuries:

- start 0
- capital: pig, fish (2nd ring), decent riverage
- good sites for second/third/fourth cities

- start 1
- capital: hillpig, drycorn, decent riverage. decent capital.
- huge tundra wasteland right in front of it
- desert to its southeast
- very little food to the north.
- this site is fucked up actually, wow! this starts gets worse and worse the more i look.

- start 2
- capital: drycorn, fish (1st ring), sheep
- good 2nd/3rd cities
- needs an early galley
- imp very good for this spot

- start 3
- triple grass sheep, dry corn
- ok sites for 2nd city, tougher after that

- start 4
- capital: TWO HAMMER PLANT, triple floodplain, grass sheep, plains cow, very good rivers
- shitloads of jungle to the south
- decent 2nd/3rd cities; things might get slower after that though

- start 5
- capital: hillpig, drycorn, no fresh water
- low trees compared to other starts
- not a lot of great 2nd city options, but ok 3rd/4th cities
- cre helps this spot immensely imho
- this spot needs to control the island to its nw, but also doesn't want to give up the choke to the east nor allow a foothold to the continent to the north. very complicated start.

- start 6
- capital: copper (forested), a single fish (first-ring), multiple jungles in capital bfc, very good riverage
- land nearby capital sucks a lot actually
- long term potential outside of capital zone is amazing however

- start 7
- capital: hillpig, fish (1st ring), cow, marble, no fresh water or rivers
- insane moai spot nearby, but bad as an early city unless cre
- WARNING: if the capital is planted on the moai, this spot is FUCKED
- not great sites for 2nd/3rd city imho
- this start wants the island to the south so bad

- start 8
- capital: wetcorn, grasshill sheep, fish (1st ring), nice rivers
- ok sites for 2nd/3rd cities
- boned after that

- start 9
- capital: marble+double stone, sheep + drycorn if settler moves onto plainshill marble
- great spots for 2nd/3rd cities
- wants Ind, heh

Some very small changes that I think would improve the map drama:

1.) give player 5 a backdoor way to get across the bay, because its way more likely player 0 will beat them to the choke because player 0's start is much faster



2.) add some floodplains to the big chokepoint desert between 0/3/8 to make that area worth fighting over




3.) swap the location of the marble and aluminum tiles, because if player 7 moves to settle on the marble where it is they'll end up orphaning two fish and ruining an amazing moai site. or, do something else, like add an extra land tile near the fish so they won't be orphaned by the marble plant. just make it so they don't want to commit seppuku on T1.




4.) player 6's capital spot turns out to be considerably slower than every other players'. like, by a ton. they only have a single fish, from a 10T workboat, and multiple jungles in their bfc, where most other players have a 6f + a 5f, or better. (some much better, e.g. see player 3). my suggestion is to get rid of the capital bfc jungle and move this pig down one tile, into the bfc. i think it's worth buffing the capital like this because otherwise it's tough to judge the spot; everything else about start 6 is awesome, but this player will have 2 cities out when others have 5. note that even with my suggested buff this is still the worst capital! note that if they yolo-move onto the banana their capital is completely without food.




5.) it is insanely important for player 1 to have a shot at controlling this area, as their starting zone is totally horrible and so their only real land will be the peninsula to the south. a plains wheat here will help anchor this position.

blah blah blah. anyways, here's my ratings, with 5 being "average", 1 being the best, 10 being the worst.

player 0: 4
player 1: 9 (10 without the suggested buff)
player 2: 1.5
player 3: 3.5
player 4: 3
player 5: 5.5 (6.5 without the suggested buff)
player 6: 1 (5 without the suggested buff)
player 7: 6 (8 without the buff)
player 8: 9
player 9: 7

this is also assuming that strat resources get fixed up.

I'd agree with this, looking at starts by themselves, but does this work out okay with neighbors then?

4 (Dan here) is in a scary place; 8 (Gav here) can also eat a lot. Makes poor 9 (2mn) stuck in a rough crowd too...

...although maybe something like this could be viable, if we then use the combo quality to adjust things too...

[Commodore]

Honestly, neighbors >>> all else. If our solution doesn't wedge OH between 2mnot4e and Gavagai, and ensure none of those three have easy snack access, we're doing it wrong.

[GermanJoey]

I'd agree with this, looking at starts by themselves, but does this work out okay with neighbors then?

Honestly, neighbors >>> all else. If our solution doesn't wedge OH between 2mnot4e and Gavagai, and ensure none of those three have easy snack access, we're doing it wrong.

In the end, if we go with my idea for adjusting a player's score based on neighbors, then neighbors end up mattering a huge deal. Let me explain.

First, here's what I've tallied so far for player ratings:




Notes below. If I misinterpreted anything, please let me know!

We had a bunch of different rating schemes, so here's how I translated stuff from into a scale from 1 to 10:

• For TBS, he ranked people into tiers; so I translated Tier 1 into a 10 and Tier 4 to be a 1, and then Tier 2 became a 7 and Tier 3 a 4. He then said Molach was a slight notch below Tier 4.
  
• Bacchus rated players from 1 to 5, so I doubled his values.
  
• Dreylin said he liked my ratings so I duplicated them.
  
• Molach ranked people in order, so I translated that from 10 to 1.
  

For neighbor ratings, Dreylin and Commodore liked the idea I had (OH/2mn/Gav/REM reduce their neighbors' skill by 1 point, Greywolf/TW/Dantski increase their neighbors' skill by 1 point). Bacchus OTOH rated players from 1 to 5 again. So, I figured 3 would be the middle. So, a 5 rating became a +2, and a 2 rating became a -1.

Most ratings were pretty consistent except for 2metraninja's, especially since some of these numbers are from before OT4E joined his team. But anyways, I just took everyone's posted rankings and averaged them, and bam there's a number for a player.

Then we have the map rankings so far. WE NEED MORE MAP RANKINGS!!! Here's how it works out with your suggested map layout:




Again, I just took the average map ranking from what's been posted so far.

Finally, I calculated how close the player's skill matched the spot with the given neighbors. The formula is this:

PLAYER_SKILL_RATING - STARTING_SPOT_RATING - sum(NEIGHBOR_RATING) = ERROR OF FIT

IMHO an error for a particular player between -1 and 1 is a really good fit. between -1.5 and 1.5 is probably still good enough because we can still adjust more later with the three leader/civ combos we give each player.

I took the player skill number from the first image and subtracted the value for their start. That's the value in column C. Then I looked at each of their neighbors, and added their values up. For overseas neighbors, I divided the neighbor factor by 2.

For example, Alhazard scored 4.67 on skill and his start, the one for player 5, scored 5.17, and therefore he's at -0.5 in column C. Negative values in column C mean that the spot's difficulty is a higher than the player's skill.

Alhazard's neighbors are REM by sea (0.75/2), Greywolf by land (-1.25), and Molach by land (-0.25). So, his value in column D is (0.75/2) + (-1.25) + (-0.25) = -1.125. Negative values in this column mean that the player has an easy neighbor set.

Finally, to calculate the error in column E, we just take C-D, -0.5 - (-1.125) = 0.625. Which is a pretty good fit for him.

Honestly, this is a pretty good fit overall. The one that really sticks out like a sore thumb is 2metra, but again the ratings of him were all over the place.

So, what do these error values tell us? Basically, positive error values translates into "be mean with giving out combos", and negative error values translate into "be nice with giving out combos." Bigger scores mean we should be meaner/nicer. For example, we should be really mean to OH and Greywolf, but pretty nice to 2metra and REM. 2metra's start maybe should be buffed a little bit too, because his error is really out there... maybe toss him a grass cow next to his marble or something.


here's the spreadsheet. I think I need your email if you want to edit it; it doesn't let me just make it editable by anyone. Or, you can just post your new/updated ratings in this thread and I'll be happy to put them in the spreadsheet for you.

https://docs.google.com/spreadsheets/d/1...sp=sharing

[GermanJoey]

If nobody wants to read that, here's the tl;dr version:

WE NEED MORE PLAYER AND ESPECIALLY MAP RATINGS!

this doesn't have to be like, super scientific or whatever. even if its something like, "i tripped and dropped my cereal on the floor this morning, and based on a divinination of the pattern of soggy frosted flakes splattered among porcelain shards I have derived the following ranking..."

really, its ok, just post some numbers! if you don't know a player well and don't feel comfortable rating them, just skip over them.

[DTG]

Tier 1: OH
Tier 2: Gavagai = 2Metra/OT4E
Tier 3: REM
Tier 4: Alhazard = ipecac > Molach
Tier 5: Dantski / Greywolf / TheWannabe <---- I am not comfortable with ranking or differentiating between any of these.

I don't know OT4E at all, but scan reading his posts in PB22 (I think it was 22) makes it clear to me that he is very good.

For the map: I would thin out some of the Forest/Hill clusters a bit. It's fucking annoying to have every Scout/Worker/Settler move capped by 1, especially around the capital, e.g.





Keep Jungle away from 2nd or 3rd city food resources, if the site is obvious, e.g.



I would be most worried about TheWannabe's start. I *think* I would flip his starting island with the island currently to the east (i.e. move the entire island, not just the capital), so that his core is a bit closer to OH and 2MN, thereby adding a sense of urgency to his situation. At present, OH is just going to get onto one of the islands to his SE, and then use this as a platform to nibble away at TW's holdings one island at a time (doubtless leading to an OH win). Or else, you could give TW a good reason (Gems?) to expand NE, and put hills on the obvious city spots to help protect his cities. Since he's a Tier 5 player, there's a real danger that he will realize he is largely isolated, feel comfortable, not push for the difficult to settle lush islands, miss the wonders needed to supercharge his economy, go into autopilot mode (= build too much infra/too few Galleys and military), and finally fall into irrelevance.

Basically, a Tier 1 player can utterly dominate with an isolated start, which is why Gawdzak did so well in PB22, but weaker players don't necessarily know how to break out, leading them to play a largely passive and unsuccessful game.

[wetbandit]

Uncaffeinated rankings

Tier 1: OH
Tier 2: Gavagai = 2Metra/OT4E
Tier 3: Molach = Alhazard = ipecac = REM
Tier 4: Dantski / Greywolf / TheWannabe

[AdrienIer]

There's something I don't get : Why don't we place the players on the map based on neighbors and available space alone, then tweak the land quality to make it match the player's quality ? Would it take that long to add a few resources to one start and delete some in others ?

[Bacchus]

I would maybe swap Ipecac and REM.

On TW's start, maybe the two southern islands should be shifted 2S, whilst maintaing galley connection with the rest of the archipelago and the northern continent.

Also, just to make sure we are not overlooking stuff, I guess a public checklist is in order, which we can go through for each player: fresh water, strat resources through the modern age, jungle placement, reasonable early city spots, access to the relevant seas, luxuries, marble/stone, etc.

[Commodore]

There's something I don't get : Why don't we place the players on the map based on neighbors and available space alone, then tweak the land quality to make it match the player's quality ? Would it take that long to add a few resources to one start and delete some in others ?

I'd support this, so some extent. Really in particular would want to make that 3-player continent Gav-OH-2mn for neighborism's sake.