Inferential Game: The Foraging (Ex-)Bandit
post by abstractapplic · 2024-11-11T16:59:42.058Z · LW · GW · 4 commentsContents
Notes None 4 comments
The encampment is empty when you awaken. This isn’t necessarily a bad sign, as you weren’t sure you’d wake up at all: there were a lot of ways your fellow bandits could have taken your impassioned impromptu speech exhorting them to “be a bit more Robin Hood about the whole thing”, and departing in the dead of night was by no means the worst. Between your continued existence, and the surprisingly generous amount of weapons and ammo which were left behind with you, you find yourself feeling rather relieved, and even a little grateful.
These feelings last until you visit the nearest town – hoping to trade bullets for food and a ticket home, so you can forget this sorry chapter of your life – and notice WANTED posters featuring your face on every wall. Your brief and understandably tense interactions with the locals clarify that you have been named as your erstwhile gang’s ringleader, that you are emphatically unwelcome in any of the shops in town, that a team of unstoppable mercenaries will be coming for you in a little under two months, and that “I assumed they were dashing desperados when I joined” is not typically considered an extenuating factor.
Your only remaining option is to forage for food and head home on foot, before the bounty hunters show up. Fortunately, your campsite was chosen for its closeness to an abundance of easy foraging sites, and you’ve already figured out how to safely and reliably preserve and prepare the ingredients sourced from them. Unfortunately, you were too busy cooking to help your fellow bandits forage[1], so you have no idea which sites yielded which yields in what quantities; also, the only viable path back has no opportunities to replenish your stockpile en route.
You have sixty days to explore and exploit the sites around the encampment, hunting[2] and gathering enough food to keep yourself alive, and stockpiling enough to reach safety before starving. Good luck!
Notes
- This game shouldn’t take more than ~10min to play, runs in-browser, and is not intended to be replayed.
- Unlike in the classic bandit problem, outcomes can be affected by chronological effects and/or earlier choices. Figuring out where and how this happens is an intended part of the challenge.
- Data Science skills might be useful here, but will not be necessary.
- You are warmly encouraged to share and compare completed game records in the comments.
- Feedback, as always, is greatly appreciated.
- ^
You’d originally thought they were asking you to be their chief, and by the time you figured out your mistake it was too awkward to back down.
- ^
You are sufficiently well-armed that nothing you hunt will be able to harm you. Indeed – you consider, feeling your revolver in its place on your belt between your other revolver and your other other revolver – you could justifiably call your current predicament a multi-armed (ex-)bandit problem.
4 comments
Comments sorted by top scores.
comment by noggin-scratcher · 2024-11-12T13:04:48.018Z · LW(p) · GW(p)
not intended to be replayed
I have flagrantly disregarded this advice in an attempt to uncover its secrets. I'm assuming there are still a bunch of patterns that remain obscure, but the ones I have picked up on allowed me to end day 60 with 5581 food just now. So I'm calling that good enough.
Rat Ruins:
Starts out rich but becomes depleted after repeat visits
Dragon Lake:
I don't think I've ever seen food here. Dragons not edible?
Goat Grove:
Good at the beginning, gradually runs down as time passes
Horse Hills:
A few random hours of each day (if there's a pattern I haven't spotted it) will return numbers in the 20s or 30s, small numbers otherwise
Tiger Forest:
Good in the last 2 or 3 hours of each day, small numbers otherwise
The rest:
Experimented with spending all day every day in any given territory - some broadly net-positive, some net-negative, but nothing seemed very exciting. Possibly they respond to more complicated conditions that I haven't yet tried
Combined strategy:
Alternate 14 hours in Goat Grove with 2 hours in Tiger Forest as a daily routine. When Goat Grove starts to drop off to single digits per hour (around day 12–14), switch to Horse Hills. At some point hit Rat Ruins for 10 hours or so.
comment by Martin Randall (martin-randall) · 2024-11-11T17:56:36.947Z · LW(p) · GW(p)
I scored 2200 ish with casual phone play including repeatedly pressing the wrong button by accident. I'm guessing better play should get someone up to 4,000 or so.
Given the setup I was sad there wasn't an explicit target or outcome in terms of how much food was needed to get home safely. I think also a more phone-friendly design would have been nice.
Thanks for making the game!
Replies from: abstractapplic↑ comment by abstractapplic · 2024-11-14T20:05:13.832Z · LW(p) · GW(p)
Given the setup I was sad there wasn't an explicit target or outcome in terms of how much food was needed to get home safely.
Good point; I've amended the game accordingly. Thank you.
comment by Yonge · 2024-11-14T00:31:24.234Z · LW(p) · GW(p)
I starte off by going to each one in turn. And then visiting those sites that yielded the best results. It rapidly became clear that some sites were better at different times of the day, and visiting one site too much could result in diminishing returns. Then I spent an entire day at each site in turn and then visited the sites that produce the most food at a given time. The relative order of visiting sites did have an affect so this didn't always work particularly well. After a bit of playing around and experimentation I ended up settling on the following cycle towards the end:
08:00 Pig Swamp
09:00 Bunny Ride
10:00 Snake Shores
11:00 Dog Valley
12:00 Dog Valley
13:00 Dog Valley
14:00 Dog Valley
15:00 Rooster Peaks
16:00 Monkey Meadow
17:00 Bunny Ridge
18:00 Pig Swamp
19:00 Pig Swamp
20:00 Tiger Forest
21:00 Tiger Forest
22:00 Tiger Forest
23:00 Tiger Forest
This got me 2761 food total.
I enjoyed this. Though I think the 10 minuite estimate is far too low. This leaves less than 1s per decision!