Weighted NonSea Gem Production (non Rare)

chrispedersen · #1 March 6th, 2009, 02:30 AM

Huh. When I edit it, it looks normal.
Anyway the distribution is attached in xls format.

Some interesting results:

1. Not even including penalties associated with farmland, you will find 20% more gems by looking in forests. And about 350% more than if you look in caves.

(Which has implications for balance in all maps using caves, like team arena).

1b. Caves suck.
1c. Mts were 2nd best.

2. Finding nature, was more than twice as likely in forests - which makes sense. But it was also 50% more likely to find water. Much more so than swamps or coasts. (Weird).

3. If you are going to look for air, more than any other type, (except holy) it pays to look with a level 3. Holy just sucks generally.

4. Death was by far the most common gem type, followed by air.

5. Nature and Astral were by far the rarest.

6. Forget looking for non rare blood sites. Forget using bowl of blood (Note to self: make bowl of blood look for sorcerous sites, instead of merely blood, and increase the cost).

7. IF you're not looking for rare - you need L4 nowhere.

Lingchih · #2 March 6th, 2009, 02:56 AM

Hmm. That's a really interesting spreadsheet. Thanks chris. It looks like different terrain provs are really much more balanced than you would think. And, yes, it looks like cave provs really suck.

OmikronWarrior · #3 March 6th, 2009, 04:29 AM

If my memory serves, don't different terrain types increase site frequency from the setting Average? If so, you're chart would be off.

cleveland · #4 March 6th, 2009, 07:49 AM

Are "Unique" sites included in your tallies? What about "Rare" (i.e. freq=2) sites?

If so, they should probably be removed.

EDIT: Also, the totals should be normalized to the number of possible sites in the province.

Agema · #5 March 6th, 2009, 09:14 AM

Firstly, many thanks for the analysis, it's very interesting.

It could be developed further. The sums for each number offer the percentage usefulness of searching a particular terrain type with a certain magic type; this could be further adapted with the basic terrain modifier. Admittedly, this could be putting into numbers what an experienced player may already roughly know by experience to an extent, but still worth knowing.

However, is it accurate to say you get far more gems from forests? I'm guessing the totals you have added up in the main body represent *sites*, not available gems, as the magic path subtotals don't add up to the gem totals at the end. On this assumption, caves have fewer possible different site types, but not necessarily fewer sites or gems. Continuing my subtotals = sites assumption, If you divide gems by the sites, all terrain types are between 1.85 and 1.90, which would mean your income per site is effectively the same in any terrain: the big modifier will be the likelihood of a site being in a province.

Finally, it might be nice to see the number of gems of each type per site path. For instance, 20 nature sites in forests might provide 18N, 2F, 6A, etc. That's a lot of database searching and number crunching so I certainly don't mean to put any onus on anyone. But it could be a useful step to optimising where you search for certain types of gems, assuming limited means.

Agema · #6 March 6th, 2009, 09:37 AM

Here's a provisional adaptation of the info, based on my inferences from the original spreadsheet.

chrispedersen · #7 March 6th, 2009, 11:25 AM

Quote:

Originally Posted by Agema

Firstly, many thanks for the analysis, it's very interesting.

It could be developed further. The sums for each number offer the percentage usefulness of searching a particular terrain type with a certain magic type; this could be further adapted with the basic terrain modifier. Admittedly, this could be putting into numbers what an experienced player may already roughly know by experience to an extent, but still worth knowing.

However, is it accurate to say you get far more gems from forests? I'm guessing the totals you have added up in the main body represent *sites*, not available gems, as the magic path subtotals don't add up to the gem totals at the end. On this assumption, caves have fewer possible different site types, but not necessarily fewer sites or gems. Continuing my subtotals = sites assumption, If you divide gems by the sites, all terrain types are between 1.85 and 1.90, which would mean your income per site is effectively the same in any terrain: the big modifier will be the likelihood of a site being in a province.

Finally, it might be nice to see the number of gems of each type per site path. For instance, 20 nature sites in forests might provide 18N, 2F, 6A, etc. That's a lot of database searching and number crunching so I certainly don't mean to put any onus on anyone. But it could be a useful step to optimising where you search for certain types of gems, assuming limited means.

cleve: As the title says its all non-rare sites. I did not thoroughly check for uniques, but cursorily I believe I eliminated those as well.

agema: No, these are gem figures being produced.

agema2: As for distribution of gems, by population type, I've already got that data as well, just thought it was a little much for people to digest.

Whoever said it should be normalized to the number of sites - is absolutely correct. However, there was a limit to the number of hours I was willing to spend last night. I'll do more later.

chrispedersen · #8 March 6th, 2009, 06:28 PM

Well the data smoothed out a lot.
I found that some common uniques had crept into the data (thank you cleveland).

After analzying for that, there is still a variation in prevalence and distribution of sites, but its much lower.

So far, caves have come in at 1.52 odd gems per site, swamps at 1.69 gems per, and forests at 1.65.

Forests: N, D,S
Swamps: D, W,S
Caves: E

Aso interestingly, so far no blood sites provide blood income.

vfb · #9 March 6th, 2009, 07:04 PM

Blood sites are not useful for slave income, they are useful for casting discount (and summons) only. When you get 60% off, it makes a big difference. Even 40% off makes a big difference.

Edi · #10 March 7th, 2009, 05:56 AM

What the chart also means is that there is a greater variety of sites available for those particular terrain types, but given that there are only 4 sites per province max, the total gem income for any given province is going to be more or less the same, all other things being equal. Which they are not, as the site frequency modifiers due to terrain kick in.

Swamps and wastes have a site frequency 10 percentage points greater than forests or mountains and 20 greater than plains.

The Looking for advice on site searching thread has some very relevant data for this analysis. It doesn't cross check terrains and paths, but it does break sites down by path and rarity and has the terrain modifiers.