Topic: Migrant numbers (Read 1929 times)

Madrigal · « **on:** December 27, 2018, 08:52:06 pm »

I hate migrants.

Specifically, I hate the giant wave of migrants that shows up in Spring 2.

According to the wiki, migrant wave size is influenced by a fort’s created wealth. So I decided to figure out how much wealth it would take to attract a given number of migrants each season, by restarting the same fort again and again, producing different amounts of wealth each time.

I did not find what I was expecting to find.

Overview:
At first I ran a few “typical” forts (file 1’s) to make sure that migrants would appear, and find out how many there would be (lots). Then I tried some meager but still functional forts (file 2’s), and had strangely uncertain results: sometimes zero migrants, and sometimes ten-ish per season, at the same level of wealth. Then one “high wealth” fort (file 3), with wealth artificially boosted by spiked wooden ball traps; it got plenty of migrants, more than the “typical” forts, but not by a lot. Then several “dirt hole” forts (file 4’s) at wealth levels below the file 2’s, trying to prevent artifacts, since I suspected they might affect the results; in these forts I again had strange results, sometimes a few migrants, sometimes none at all. Then several “modest” forts (file 5’s) at wealth levels mainly between the file 2’s and the file 1’s, again preventing artifacts... and again getting sometimes just a few migrants, sometimes none at all, even at wealth higher than the file 1’s, until I tried repeating a couple of these forts with high item creation, which triggered both artifacts and higher migrant numbers. Finally, the file 6’s all had moderate-to-high item creation, enough to trigger at least one artifact, and consistently saw higher migrant numbers.

RESULTS SUMMARIES.

Spoiler (click to show/hide)

Findings:
-- Higher created wealth (presumably since the previous caravan report) did generally attract more migrants. However, a specific amount of created wealth did not reliably attract a specific number of migrants (compare files 1.2, 1.2b with 5.8, 5.8b; compare files 5.4, 6.1, 6.4, 6.5, 6.6). Forts with the most migrants tended to be forts with BOTH high wealth AND lots of jobs/items. So, items/jobs might be as relevant as (or more relevant than) created wealth.
-- Forts with very few created items (to prevent artifacts) often failed to get migrants even if the created wealth was overall fairly high (see all the file 4’s and file 5’s). Also, in these forts, migrants were more likely to appear if the caravan left before autumn migrants showed up (that is, the dwarf population at the time was smaller, meaning higher wealth/jobs/items per dwarf): compare files 4.1, 4.3, 4.4 with 4.2; compare file 5.6 with 5.5 and 5.7. So, per-dwarf values might matter instead of total values. However, a specific amount of created wealth per dwarf did not reliably attract a specific number of migrants (compare files 4.2 and 5.1 year2 with files 5.6 and 5.1 year1; compare files 5.4 and 6.4). Possibly these variations are due to differences in the number of jobs done or items created; I usually didn’t track precise numbers of these. Or possibly the per-unit values are relevant (instead of per-dwarf), but I didn’t gather that data.
-- There were two different migrant patterns: either roughly equal numbers of migrants each season (see files 2b, 4.2, 5.1, 5.4, 6.1, 6.6), or noticeably higher numbers in spring (see files 1.2, 1.2b, 1.3, 5.8b, 5.9b, 6, 6.2-6.5). This might indicate that there are two or more factors affecting migrant numbers, instead of just one.
-- It was surprising to me that a relatively small number of created items seemed to attract migrants so much: files 6.1, 6.4, 6.5, 6.6 all got more migrants than I expected (10-ish or more per season), based on something like a dozen rocks made into blocks and another 100ish loose rocks lying around.

A tentative hypothesis: one factor (possibly created wealth, possibly per dwarf or per unit) provides a baseline number of migrants every season, but some other factor (such as number of jobs done, or number of items produced, or jobs/items per unit, or...) provides extra migrants, many/most of whom show up in spring. If true, this would explain why the “low number of items” forts (file 4’s and 5’s) generally got similar numbers of migrants each season (if any), while the “high number of items” forts (1.2, 1.2b, 5.8b, 6) often had the “spring bulge” and higher migrant numbers overall. Unfortunately, supporting this hypothesis would require more careful tracking of jobs/items than I’m willing to do, so I’m ending this here.

I suspect that no one other than me is terribly interested in this issue, given how easy it is to set the population caps in d_init.txt. But if others can confirm that “jobs and/or item production help attract migrants” or “per-dwarf or per-unit values are relevant,” it might be worth updating the wiki.

Sarmatian123 · « **Reply #1 on:** December 28, 2018, 02:37:28 pm »

Interesting finds. Thank you. It was an interesting reading. Indeed many factors affect how many Dwarfs migrate.

-Fortress wealth suppose to be one of those factors.
-How many are killed in fortress (even animals I think).
-Total available population of this civilization in world.

Dead civilizations particularly severely nerf down migration waves.
10 hunters busy exterminating animal populations also affect migrant waves.

I would guess wealth affect civilization's already existing settlements. Acts as attractor of sorts. Not only to Mega Beasts. So, if there are some same amount of Dwarven settlements with a given population, then recruitment drive maybe is impacting only those Dwarves as potential candidates for migration?

My current cap on population is 150/175. My former 100/120 was not working with new features and you can't obtain metropolis status without 140 Dwarves and get king to move in or have hillock farms around your fortress. Hillock farms is something I wish to check out, if my Fortress won't die from mass-depression first, that is.

feelotraveller · « **Reply #2 on:** December 28, 2018, 09:50:46 pm »

I'm one of those uninterested d_init tweakers.

But your mention of the 'job/item production' factor along with the 'spring bulge' makes me wonder if there is a connection here with the fortress title (rank). See the second section here: http://dwarffortresswiki.org/index.php/DF2014:Fortress. If you still have the saves from the above tests, or are willing to go another round, it might be worth having a look to see if/how this hunch plays out.

Anyway good testing - I always find that fascinating.

Madrigal · « **Reply #3 on:** December 29, 2018, 01:33:47 pm »

Sarmatian:
Yes, indeed, deaths, and the civilization's population and other settlements, are factors that should affect migration; I was generally trying to eliminate these variables to examine only the effect of fort wealth, but others are free to do those experiments, if they'd like!

Feelotraveller:
I did keep the saves, one for each file, but I'm a little eensy weensy bit bored with this experiment, so I'm not really willing to keep going, beyond keeping track of jobs and population in my future forts.

I suspected the same thing about diverse jobs and fort rank, but my main problem was that I'm not absolutely sure of what, exactly, counts as a job of each type. For example, it's obvious that turning a log into a bed is a "wood job", but what about cutting down a tree? Or burning a log into charcoal or ash? Is smelting ore a "metal job" or a "stone job"? Is processing pig tail into thread a "craft job"? If there is a list, somewhere, of exactly what activity counts as exactly what kind of job, I'd appreciate being pointed in that direction.

feelotraveller · « **Reply #4 on:** December 29, 2018, 06:53:19 pm »

I think that the type of job is defined by the associated Labor category (see section in http://dwarffortresswiki.org/index.php/DF2014:Labor). I'm pretty sure woodcutting is a wood job, smelting is a metal job and that pig-tail processing is not a craft job.

But we don't necessarily need to know this here. The quick way to confirm through observation is to look at the 'z' screen with the arrival of the dwarfen caravan. In the top left will be the rank (and name) of the fortress. Then next spring, as the migrants are announced (or not) check again the 'z' screen. If consistently when the rank increases there is a bulge of migrants, and not otherwise, we have confirmation. It would be nice to get some estimate of the strength/power of this factor - how much bulge with fortress rank-up?

It is true that knowing the job types would enable manipulating the situation to provide rank increase events, or not, for testing purposes. I think from the above link that we can be pretty sure of the relevant job types for all except 'food'. (gem jobs = jewelry)

Someone else (a modder perhaps) might know this in more detail...

Madrigal · « **Reply #5 on:** June 11, 2019, 09:31:47 pm »

It's been a while, but I finally did a brief follow-up check on my little experiment.

Feelotraveller, your idea that a change to the fortress’s title (i.e., outpost, hamlet, village, town...) might attract extra migrants in spring seemed like a great idea, especially since it would explain the weird spring-only migrant situation from file 1.1b. I wish I’d been noting down the fort titles throughout! However, I went back to my files and checked the fort titles for the files that I only ran until year 2 or thereabouts. Unfortunately, the idea doesn’t seem to be supported by the (very small amount of) evidence.

File 5.8, an interim save from winter 1: I was planning to run this file until I saw the fort’s title change. But as checked in early winter 1 it was already a hamlet. There were no migrants in year 2.
File 6.4: There was a “spring bulge” in spring 2, but as checked in mid summer 2 it was still just an outpost.
File 6.5: A “spring bulge” in spring 2, and as checked in early summer 2 it was a hamlet.
File 6.6: If there was a “spring bulge” in spring 2, it was very small. As checked in mid summer 2, it was a hamlet.

Madrigal · « **Reply #6 on:** September 08, 2019, 06:27:16 pm »

Finally, some clear results!

In this investigation, File 7, I changed the amount of mining/jobs while keeping population steady.

All of the File 7 forts are restarted from the same save file (same fort as above, Kulalineth): 1st day of autumn, population 18, minimal fort (minimal furniture/workshops/farming/booze production, and just over 500 tiles mined out, plus 40ish tiles channeled from the surface and floored over), and the autumn migrants arrive on 26 midautumn (after the caravan leaves). Note that all jobs done in spring/summer of the first year are the same in every file; the only difference between the files is the jobs/mining starting in early autumn (listed below). Fort wealth, as of the time the merchants announce they’re leaving, is also reported. Strange moods are listed if they occur. Files are listed in the order of number of jobs done (considering each mined tile to be “one job”).

Files with no migrants:

Spoiler (click to show/hide)

Files with migrants:

Spoiler (click to show/hide)

It looks like mining and jobs attract migrants. Specifically, about 100 jobs needed to be done (and/or tiles needed to be mined out) -- on top of the jobs & mining done in spring/summer -- before migrants would appear in this instance of Kulalineth in the second year.

I’m not sure if there’s anything special about mining itself; it might just count as a job/task/activity/whatever that can be done quickly.

I was surprised to see that file 7H had no migrants; 320 blocks produced and no migrants seems to show that migrants are not noticeably attracted by item production.

The effect of fort wealth:
-- Compare 7R and 7H (both with over 4500 wealth reported and no migrants) to 7L, 7I, 7J, 7K (all with under 4000 wealth reported, all with migrants, and sometimes fairly high migrant numbers, too).
-- Compare 7Z (high wealth, single digit migrants) with 7K, 7S, 7T, 7U, 7V (lower wealth, double digit migrants).
-- But compare 7Z and 7W. Same job numbers, but 7Z had higher wealth, and also had higher numbers of migrants in the spring & summer... but 7W had more autumn migrants, so the total population as of the summer2 migrant wave was roughly the same for both. In all, I’m not sure what to make of this comparison.

===
So! Based on the File 7 findings, I set up a File 8 fort to maximize mining (Kulalineth again, restarted from the first spring): a minimal fort, making a half dozen picks in the first summer, making virtually all adult dwarves into miners, and designating roughly 5k tiles for mining in addition to the 370ish tiles mined out early on for living space.

File 8 results:
7 starting dwarves + 5 summer migrants + 4 autumn migrants, pop 16. 16 midautumn, “merchants embarked,” wealth 14694, 4503+370 tiles mined at moment of announcement. Spring: 37 migrants, pop 53, still outpost. Strange mood. Summer 1 birth, 15 migrants, pop 69.

Compare this high number of migrants to File 3 as reported above. File 8 achieved the same high migrant numbers as File 3, but with half the reported wealth. At this point I’m pretty comfortable declaring that migrant numbers are strongly affected by jobs/mining. Since jobs/mining will inevitably increase fort wealth, this could mean that wealth is only correlated with migration.

Also, File 8’s migrant numbers were high but not insanely high, even though the amount of mining was drastically higher than anything in File 7. There might be diminishing returns, or some other factor (perhaps the lack of job variety) might be affecting the results. I've pretty much satisfied my curiosity so I probably will not be investigating further, but you never know.

delphonso · « **Reply #7 on:** September 10, 2019, 03:55:36 am »

Very interesting. Thanks for the hard work!

Crabs · « **Reply #8 on:** September 10, 2019, 04:05:49 am »

Quote from: Madrigal on December 27, 2018, 08:52:06 pm

I hate migrants.

Same.
Loljk. Thanks for the effort! A true man of science!

News:

Author Topic: Migrant numbers (Read 1929 times)

Madrigal

Migrant numbers

Sarmatian123

Re: Migrant numbers

feelotraveller

Re: Migrant numbers

Madrigal

Re: Migrant numbers

feelotraveller

Re: Migrant numbers

Madrigal

Re: Migrant numbers

Madrigal

Re: Migrant numbers

delphonso

Re: Migrant numbers

Crabs

Re: Migrant numbers