Posts

D&D.Sci(-fi): Colonizing the SuperHyperSphere [Evaluation and Ruleset] 2024-01-22T19:20:05.001Z
D&D.Sci(-fi): Colonizing the SuperHyperSphere 2024-01-12T23:36:54.248Z
Fifty Flips 2023-10-01T15:30:43.268Z
Some 2-4-6 problems 2023-03-28T06:32:02.946Z
Bayesian Scenario: Snipers & Soldiers 2023-02-26T21:48:00.788Z
Durkon, an open-source tool for Inherently Interpretable Modelling 2022-12-24T01:49:58.684Z
D&D.Sci December 2022 Evaluation and Ruleset 2022-12-12T21:21:08.781Z
D&D.Sci December 2022: The Boojumologist 2022-12-02T23:39:49.398Z
D&D.Sci September 2022 Evaluation and Ruleset 2022-09-26T22:19:01.415Z
D&D.Sci September 2022: The Allocation Helm 2022-09-16T23:10:23.364Z
My Opportunity Costs 2022-07-10T10:14:58.827Z
D&D.Sci June 2022 Evaluation and Ruleset 2022-06-13T10:31:25.447Z
D&D.Sci June 2022: A Goddess Tried To Reincarnate Me Into A Fantasy World, But I Insisted On Using Data Science To Select An Optimal Combination Of Cheat Skills! 2022-06-04T01:28:18.301Z
Clem's Memo 2022-04-16T11:59:55.704Z
How I repeatedly failed to use Tobit modelling on censored data 2022-04-02T18:10:42.063Z
Lesson Plan: Biases in Quantity Estimation 2022-03-26T00:23:17.822Z
D&D.Sci August 2021 Evaluation and Ruleset 2021-08-23T22:49:20.528Z
D&D.Sci August 2021: The Oracle and the Monk 2021-08-13T22:36:38.572Z
D&D.Sci(-Fi) June 2021 Evaluation and Ruleset 2021-06-29T21:02:20.072Z
D&D.Sci(-Fi) June 2021: The Duel with Earwax 2021-06-22T11:48:44.718Z
Does anyone have any Data Sidequests? 2021-06-11T23:40:09.844Z
A.D&D.Sci May 2021 Evaluation and Ruleset 2021-05-24T16:25:13.704Z
A.D&D.Sci May 2021: Interdimensional Monster Carcass Auction 2021-05-17T15:54:28.974Z
D&D.Sci May 2021 Evaluation and Ruleset 2021-05-14T11:37:14.328Z
D&D.Sci May 2021: Monster Carcass Auction 2021-05-07T19:33:19.920Z
D&D.Sci April 2021 Evaluation and Ruleset 2021-04-19T13:26:58.278Z
D&D.Sci April 2021: Voyages of the Gray Swan 2021-04-12T18:23:11.674Z
D&D.Sci III Evaluation and Ruleset 2021-03-08T23:01:36.833Z
D&D.Sci III: Mancer Matchups 2021-03-05T19:07:17.473Z
D&D.Sci II Evaluation and Ruleset 2021-01-17T16:58:40.087Z
D&D.Sci II: The Sorceror's Personal Shopper 2021-01-12T01:38:44.168Z
D&D.Sci Evaluation and Ruleset 2020-12-12T15:00:20.984Z
D&D.Sci 2020-12-05T23:26:40.934Z
Model Depth as Panacea and Obfuscator 2020-11-09T00:02:03.297Z
Case Study II 2018-09-30T00:37:32.974Z
Applying Bayes to an incompletely specified sample space 2018-07-29T17:33:53.978Z
Excessive EDA Effortposting 2018-06-03T19:17:22.595Z

Comments

Comment by abstractapplic on One-shot strategy games? · 2024-03-11T11:41:16.866Z · LW · GW

Disrecommending Slay The Spire. While it's a great game and it fits the rest of your criteria like a glove, it has very little hidden information in a practical sense (one of the more innovative things about it is that you can almost always see what the enemy will do next turn), and as such has basically no places where explore/exploit tradeoffs and VOI calculations would be relevant (I assume that this isn't a negotiable part of what you're asking for; if not, yeah I also recommend StS).

Comment by abstractapplic on One-shot strategy games? · 2024-03-11T11:35:29.491Z · LW · GW

Tentative recommendation of Slipways; the VOI part isn't as central as I suspect you'd like, but sending out probes sure does cost time and money you could use for settling planets and forming trade routes; and while it's easy enough to survive to the end of your term, it gives what you're asking for if you choose to define 'victory' as 'get 5+ stars on Tough'.

Comment by abstractapplic on Lsusr's Rationality Dojo · 2024-02-15T20:42:56.347Z · LW · GW

The objective of rationality is to become right instead of wrong.

 

I think this is technically false, in a subtle but important way. If I gained [knowledge of whether every six-digit number is prime] in exchange for [knowledge of whether wandering out into open traffic is a good idea], I'd have gleaned a net 899999 bits of right-ness, but it still wouldn't have been a worthwhile deal, or made me more rational in any practical sense. The missing gears are becoming right about important && relevant things, bothering to apply that knowledge, and - conditional on applying it at all - applying it well.

I think this project is good (Like, unusually good! It's a step forward! I enjoyed it, and I commend you for your service to the Cause!), but I notice a lack of emphasis on changing actions vs changing minds, both in this post and in the videos I watched, and I want to make sure you've noticed that too.

(And yes, I do recognize the irony of me pointing out a true thing about [pointing out true things without having an associated practical outcome] without having an associated practical outcome. Still think it's worth saying!)

Comment by abstractapplic on D&D.Sci(-fi): Colonizing the SuperHyperSphere [Evaluation and Ruleset] · 2024-02-05T22:49:00.726Z · LW · GW

Sorry about that, reality got in the way; also, ended up scrapping my concept for the next one and my backup concept for it; no idea when it'll end up actually made (not necessarily this month), except that I plan to release on a Friday to do the standard "10 days with a choice of weekend" thing.

Comment by abstractapplic on D&D.Sci(-fi): Colonizing the SuperHyperSphere [Evaluation and Ruleset] · 2024-01-23T14:50:29.666Z · LW · GW

Damn! Mea culpa; I'll edit the original post so anyone going through the archives won't have the same problem.

Comment by abstractapplic on How I repeatedly failed to use Tobit modelling on censored data · 2024-01-21T02:11:41.329Z · LW · GW

Also, strong-upvoted for asking "so, with X years of hindsight, how did this pan out?" on an old post. More people should do that.

Comment by abstractapplic on How I repeatedly failed to use Tobit modelling on censored data · 2024-01-21T02:10:17.422Z · LW · GW

Before circumstances let me answer that question, the client got bought out by a bigger company, which was (and is) a lot more cagey about both hiring contractors and sharing internal details with outsiders; last I heard, the client's absorbed remnants are still sometimes using my modelling approach, but I have no idea how much they're using it, how much they're relying on it, or to what extent it's benefiting them.

Comment by abstractapplic on D&D.Sci(-fi): Colonizing the SuperHyperSphere · 2024-01-14T12:48:08.129Z · LW · GW

There are no time effects in the data; past trends can in generality be assumed to exist in the present.

(Good question!)

Comment by abstractapplic on D&D.Sci(-fi): Colonizing the SuperHyperSphere · 2024-01-14T12:17:49.112Z · LW · GW

The same way it does everything: in a weird, non-Euclidean manner which defies human intuition.

Comment by abstractapplic on Bounty: Diverse hard tasks for LLM agents · 2023-12-17T23:51:36.892Z · LW · GW

For the unreleased challenge, b) isn't for sale: making something intended to (eventually) be played by humans on LW and then using it solely as LLM-fodder would just be too sad. And I'm guessing you wouldn't want a) without b); if so, so much for that.

. . . if the "it must never be released to the public internet" constraint really is that stringent, I might be better advised to make D&D.Sci-style puzzles specifically for your purposes. The following questions then become relevant:

.How closely am I allowed to copy existing work? (This gets easier the more I can base it on something I've already done.)

.How many challenges are you likely to want, and how similar can they be to each other? (Half the difficulty on my end would be getting used to the requirements, format etc; I'd be more inclined to try this if I knew I could get paid for many challenges built along similar lines.)

.Is there a deadline? (When are you likely to no longer need challenges like this?) (Conversely, would I get anything extra for delivering a challenge within the next week or so?)

Comment by abstractapplic on Bounty: Diverse hard tasks for LLM agents · 2023-12-17T13:04:09.145Z · LW · GW

This seems like a natural fit for D&D.Sci games. All the ones I made are public domain, so you can use them freely (and I bet the other people who made some would give you permission if you asked them nicely), they've been publicly played by clever humans with a variety of skill levels and associated outcomes, and they're obscure enough that I doubt an LLM would have memorized the solutions (and if not you could tweak the names and data-generation hyperparameters to flatfoot them).

. . . I happen to have a completed-but-unreleased D&D.Sci game, which I was planning to put on LW early next month, after everyone got back from their holidays. Would it be helpful if I sent it to you and delayed the release until Feb, so you and yours could let LLMs try it first?

Comment by abstractapplic on New LessWrong feature: Dialogue Matching · 2023-12-11T00:58:23.374Z · LW · GW

I am in literally the exact same situation, and think your proposed remedy makes sense.

Comment by abstractapplic on A Socratic dialogue with my student · 2023-12-06T14:40:47.159Z · LW · GW

I haven't eaten meat in months.

 

Completely orthogonal to any of the more interesting points you were trying to make, but: it looks like you might be going vegan in an unsystematic way. I heard this gives people severe permanent disabilities, in ways that are trivial to dodge once you know what they are. (I realize you've probably already addressed this, but thought I'd err on the side of caution and nag you anyway.)

Comment by abstractapplic on Fifty Flips · 2023-10-02T03:19:49.165Z · LW · GW

>You link to index C twice, rather than linking to index D. 

Whoops! Fixed now, thank you.

Comment by abstractapplic on D&D.Sci 5E: Return of the League of Defenders Evaluation & Ruleset · 2023-06-10T12:42:12.578Z · LW · GW

Reflections on my performance:

I failed to stick the landing for PVE; looking at gjm’s work, it seems like what I was most missing was feature-engineering while/before building ML models. I’ll know better next time.

For PVP, I did much better. My strategy was guessing (correctly, as it turned out) that everyone else would include a Professor, noticing that they’re weak to Javelineers, and making sure to include one as my backmidline.

Reflections on the challenge:

I really appreciated this challenge, largely because I got to use it as an excuse to teach myself to build Neural Nets, and try out an Interpretability idea I had (this went nowhere, but at least failed definitively/interestingly).

I have no criticisms, or at least none which don’t double as compliments. The ruleset was complicated and unwieldy, increasing the rarity of “aha!” moments and natural stopping points during analysis, and making it hard to get an intuitive sense of how a given matchup would shake out (even after the rules were revealed) . . . but that’s exactly what made it such a useful testing ground, and such valuable preparation for real-world problems.

Comment by abstractapplic on D&D.Sci 5E: Return of the League of Defenders · 2023-05-31T00:40:23.133Z · LW · GW

Just recording for posterity that yes, I have noticed that

Rangers are unusually good at handling Samurai, so it might make sense to have one on my PVE team.

However, I've also noticed that

Rangers are unusually BAD at handling Felons, to a similar or greater degree.

As such,

I think it makes more sense to keep Pyro Professor as my mid-range heavy-hitter in PVE.

(. . . to my surprise, this seems to be the only bit of hero-specific rock-paper-scissors that's relevant to the PVE challenge. I suspect I'm missing something here.)

Comment by abstractapplic on D&D.Sci 5E: Return of the League of Defenders · 2023-05-27T16:24:02.634Z · LW · GW

Threw XGBoost at the problem and asked it about every possible matchup with FRS; it seems to think

my non-ML-based pick is either optimal or close-to-optimal for countering that lineup.

(I'm still wary of using ML on a problem instead of thinking things through, but if it confirms the answer I got by thinking things through, that's pretty reassuring.)

Therefore, I've decided

to keep HLP as my PVE team.

And I've DM'd aphyer my PVP selection.

Comment by abstractapplic on D&D.Sci 5E: Return of the League of Defenders · 2023-05-27T14:26:13.097Z · LW · GW

My main finding thus far:

There's a single standard archetype which explains all the most successful teams. It goes like this: [someone powerful from the MPR cluster, ideally P], [a frontman, selected from GLS], [someone long-ranged, selected from CHJ]. In other words, this one is all about getting a good range of effective ranges in your team.

My tentative PVE submission is therefore:

Hurler, Legionary, Professor

However:

  • I'm pretty sure there's some second-order rock-paper-scissors stuff going on that I'm not accounting for: Rangers seem better than Professors at beating Samurai in particular, Marauders seem to have a similar speciality when fighting Tyrants, Duelists/Bandits beat Amazons/Wizards beat Legionaries/Golems beat Duelists/Bandits
  • I haven't looked into how a bunch of strong/sturdy/snipey trios behave facing off against each other, which is relevant both because the PVE enemy is that kind of trio and because the PVP arena will probably be full of them.
  • Based on my research so far, I can't rule out that there's some secondary archetype which sucks in general but acts as a magic bullet against strong/sturdy/snipey trios in particular.
  • I have a stupidly ambitious ML thing I want to use this challenge as an excuse to (try to) do.

So it'll take me a while to decide on my PVP allocation, and I'm reserving the right to change my PVE one.

Comment by abstractapplic on Some 2-4-6 problems · 2023-03-28T12:15:25.186Z · LW · GW

Well, that's embarrassing. Fixed now; thank you.

Comment by abstractapplic on [S] D&D.Sci: All the D8a. Allllllll of it. Evaluation and Ruleset · 2023-02-28T00:48:20.430Z · LW · GW

Reflections x3 combo:

Just realized this could have been a perfect opportunity to show off that modelling library I built, except:

A) I didn't have access to the processing power I'd need to make it work well on a dataset of this size.

B) I was still thinking in terms of "what party archetype predicts success", when "what party archetype predicts failure" would have been more enlightening. Or in other words . . .

. . . I forgot to flip the problem turn-ways.

Comment by abstractapplic on [S] D&D.Sci: All the D8a. Allllllll of it. Evaluation and Ruleset · 2023-02-28T00:38:56.730Z · LW · GW

Reflections on my performance:

This stings my pride a little; I console myself with the fact that my "optimize conditional on Space and Life" allocation got a 64.7% success rate.

If I'd allocated more time, I would have tried a wider range of ML algorithms on this dataset, instead of just throwing XGBoost at it. I'm . . . not actually sure if that would have helped; in hindsight, trying the same algorithms on different subsets ("what if I built a model on only the 4-player games?") and/or doing more by-hand analysis ("is Princeliness like Voidliness, and if so, what does that mean?") might have provided better results.

Reflections on the challenge:

I found this one hard to get started with because it had a de facto 144 explanatory columns ("does this party include a [Class] of [Aspect]?") along with its 1.4m rows, and the effects of each column was mediated by the effects of each other column. This made it difficult - and computationally intensive! - to figure out anything about what classpect combinations affect the outcome.

That said, I appreciated this scenario. The premise was fun, the writing was well-executed, and the challenge was fair. Also, it served as a much-needed proof-by-example that "train one ML model, then optimize over inputs" isn't a perfect skeleton key for solving problems shaped like this. If it was a little obtuse on top of that . . . well, I can chalk that up to realism.

Comment by abstractapplic on Bayesian Scenario: Snipers & Soldiers · 2023-02-27T23:19:35.591Z · LW · GW

The jankiness here is deliberate (which doesn't preclude it from being a mistake). My class on Bayesianism is intended to also be a class on the limitations thereof: that it fails when you haven't mapped out the entire sample space, that it doesn't apply 'cleanly' to any but the most idealised use cases, and that once you've calculated everything out you'll still be left with irreducible judgement calls.

Comment by abstractapplic on Bayesian Scenario: Snipers & Soldiers · 2023-02-27T23:10:46.897Z · LW · GW

(I have the "show P(sniper)" feature always enabled to "train" my neural network on this data, rather than trying to calculate this in my head)

That's among the intended use cases; I'm pleased to see someone thought of it independently.

Comment by abstractapplic on [S] D&D.Sci: All the D8a. Allllllll of it. · 2023-02-24T11:17:24.247Z · LW · GW

If it helps, I for one am completely okay with you taking the weekend.

Comment by abstractapplic on [S] D&D.Sci: All the D8a. Allllllll of it. · 2023-02-17T02:37:38.292Z · LW · GW

I used the python package Pandas.

(I also tried Excel, but the dataset was too large to load everything in. In retrospect, I realize I could have just loaded in the first million rows - 2/3 of the dataset, more than enough to get statistically significant results from - and analyzed that, possibly keeping the remaining ~400k rows as a testing set.)

Comment by abstractapplic on [S] D&D.Sci: All the D8a. Allllllll of it. · 2023-02-14T15:17:02.749Z · LW · GW

My solution for winrate maximization:

Add a Page of Mind and a Seer of Void. (this should get us slightly better than 50% chance of success)

My solution conditional on the new universe having both Space and Life (I think Time, Space and Life are prerequisites for a universe I'd like):

Add a Prince of Space and a Sylph of Life; if the gender situation doesn't line up with that, replace the Prince with an Heir and/or replace the Sylph with a Page. (this should get us slightly worse than 50% chance of success)

My attempt at ranking the party members, based on change in predicted winrate if they're removed from the team:

The Knight of Blood is the clear MVP, followed by the Bard of Rage and the Seer of Mind. The Prince of Hope and Mage of Doom are actively harmful, and the party would be better off if something were to happen to them (for example, if they took a tumble down some stairs); the Witch of Life and the Page of Breath might also be net-negative but if so it's pretty marginal.

My advice to other players:

If you happen to be both evil and extremely gullible, play solo as a Seer of Void; otherwise, play solo as a Page of Hope.

Misc. findings:

  • Average winrate is ~1/3.
  • Very small teams and very large teams both have an advantage over medium teams; 4 is literally the worst team size.
  • There are six classpects which invariably fail a solo run; three are "X of Void", and three are "Prince of Y"; however, "Prince of Void" might be the most robustly useful archetype available. This seems like a strong hint to the generating function, but I haven't found any other perfect regularities that would help me interpret this one.
  • The generating function is impressively hard to characterize; there are very few things that don't happen. There are synergies between classes (Pages work well with Heirs), synergies between aspects (Space players work well with Hope players), synergies which transcend players (Pages work well with Void players, but Pages of Void are bad news), effects which vary with team size (most classpects which work well in single-player games are mediocre in 12-player games and vice-versa) . . . I eventually gave up on analysis and just did the "build an ML model and use it to trial solutions" thing.
Comment by abstractapplic on [deleted post] 2023-01-13T10:24:35.250Z

I just checked and while the other answers are perfect, math.log(2)**math.exp(2) is 0.06665771193088375. ChatGPT is off by almost an order of magnitude when given a quantitative question it can't look up in its training data.

Comment by abstractapplic on [deleted post] 2022-12-23T12:01:48.451Z

Thanks for putting in the time to make sense of my cryptic and didactic ranting.

You don't specify exactly how this second function can vary, whether it also has a few parameters or one parameter or many parameters?

Segmented linear regression usually does the trick. There's only one input, and I've never seen discontinuities be necessary when applying this method, so only a few segments (<10) are needed.

I didn't specify this because almost any regression algorithm would work and be interpretable, so readers can do whatever is most convenient to them.

Your approach of first optimizing f and then optimizing g, and then taking  g ∘ f as your final model has the obvious alternative of directly optimizing g ∘ f with all parameters of each function optimized together.

What I actually do is optimize f until returns diminish, then optimize f and g together. I suggested "f then g" instead of "f then f&g" because it achieves most of the same benefit and I thought most readers would find it easier to apply.

(I don't optimize f&g together from the outset because doing things that way ends up giving g an unindicatively large impact on predictions.)

is it really simpler than what you would otherwise have used for f?

Sometimes. Sometimes it isn't. It depends how wrong the linkage is.

If there are multiple input variables I'm not sure I would conceptualize this as correcting the linkage, since it's correcting the overall output and not specifically the relationship with any one input variable?

I would. When the linkage is wrong - like when you use an additive model on a multiplicative problem - models either systematically mis-estimate their extreme predictions or add unnecessary complexity in the form of interactions between features.

I often work in a regression modelling context where model interpretability is at a premium, and where the optimal linkage is almost but not quite multiplicative: that is, if you fit a simple multiplicative model, you'll be mostly right but your higher predictions will be systematically too low.

The conventional way to correct for this is to add lots of complex interactions between features: "when X12 and X34 and X55 all take their Y-maximizing values, increase Y a bit more than you would otherwise have done", repeated for various combinations of Xes. This 'works' but makes the model much less interpretable, and requires more data to do correctly.

Comment by abstractapplic on How do I start a programming career in the West? · 2022-11-26T02:06:16.433Z · LW · GW

There was a similar question a few months back; you may find the answers there helpful.

Comment by abstractapplic on D&D.Sci September 2022 Evaluation and Ruleset · 2022-09-27T23:08:29.889Z · LW · GW

Nope. (Though since both that game and this one are weird administration-centric takes on Harry-Potter-style magical schools, I imagine there may have been some convergent evolution.)

Comment by abstractapplic on D&D.Sci September 2022: The Allocation Helm · 2022-09-17T22:25:26.580Z · LW · GW

Good catch; fixed now; thank you.

Comment by abstractapplic on D&D.Sci September 2022: The Allocation Helm · 2022-09-17T02:24:42.939Z · LW · GW

It was, though fortunately that was just the random Houses they would have been Allocated to, and as such provides no meaningful information. Still, I've updated the file to not have that column; thank you.

Comment by abstractapplic on D&D.Sci September 2022: The Allocation Helm · 2022-09-17T02:22:53.223Z · LW · GW

(2)

Comment by abstractapplic on Supposing Europe is headed for a serious energy crisis this winter, what can/should one do as an individual to prepare? · 2022-09-01T13:22:10.294Z · LW · GW

Buy battery packs for charging phones so you can stay connected during a local blackout.

Comment by abstractapplic on How do you get a job as a software developer? · 2022-08-18T18:41:26.377Z · LW · GW

Wait. As . . . a software developer? Not as a Data Scientist, even though you have experience with ML?

At least as far as I know, Data work is better paid, uses more LessWrong-ish skills, and (crucially) is more of a frontier situation: Software ate the world a while ago, but Data is still chewing, so there's been much less time for credentialism to seep in.

(I'm from the UK, and it's been a few years since I did a 'normal' jobhunt, so I could be wrong about this as it applies today and on your side of the Atlantic. But even taking that into account, I notice I'm still surprised.)

Comment by abstractapplic on Dwarves & D.Sci: Data Fortress Evaluation & Ruleset · 2022-08-16T17:37:57.583Z · LW · GW

I'm curious as to what exactly you found there.

Briefly: I told my learner "assume there are two sources of income for Light Forest forts; assume they are log-linked functions of the data provided with no interactions between features; characterize these income sources."

The output graphs, properly interpreted, said back:

  • The larger source of income benefits greatly from Miners, benefits from the presence of every ore (especially Haematite), likes coal, and benefits from having one Smith.
  • The smaller source of income benefits from Woodcutters, benefits from having two (but not more) Warriors, hard-requires at least one Woodcutter and Warrior in order to be viable, actively dislikes Coal, doesn't care about ores (except Copper for some reason), and strongly benefits from Crafters.

(In reviewing my graphs in retrospect I also see a small bump in performance for both sources associated with having exactly one Brewer; I missed that the first time because it looked like noise and I'd assumed Brewers only mattered to the survival half of the challenge.)

This wasn't 100% right, and missed some important detail, but given the bad assumptions I built it on - an additive model with a lot of interactions sprinkled on top would have been a better match - I'm pleasantly surprised by how closely it matches (a valid interpretation of) ground truth.

Comment by abstractapplic on Dwarves & D.Sci: Data Fortress Evaluation & Ruleset · 2022-08-16T13:22:21.845Z · LW · GW

Reflections on my attempt:

It looks like I was basically right. I could have done slightly better by looking more closely at interactions between features, ore types especially; still, I (comfortably) survived and (barely) proved my point to the King, so I'm happy with the outcome I got.

(I'm also very pleased by the fact that I picked up on the ore-based-vs-wood-based distinction; or, rather, that the ML library I've been building automatically picked up on it. Looks like my homebaked interpretability tools work outside their usual contexts!)

Reflections on the challenge:

Another excellent entry, and a hard act to follow. The jokes landed, the premise was fun but coherent, and the scenario was challenging yet tractable.

Having multiple quantitative success metrics was a fascinating choice. To be honest, I think there was some missed potential here; if the best strategy wasn't simultaneously survival-optimal and money-optimal, there could have been some interesting tension from players deciding their blood-to-treasure exchange rates. I'll have to try and work something like that into a future game.

Comment by abstractapplic on Dwarves & D.Sci: Data Fortress · 2022-08-13T21:14:26.689Z · LW · GW

My allocations:

4x Miner, 2x Woodcutter, 2x Warrior, 2x Crafter, 1x Brewer, 1x Farmer, 1x Smith

The handful of (dubious) insights that no-one seems to have had yet, which motivate the (slight) differences between this setup and everyone else's:

  • We have enough data that it makes sense to filter out everything that isn't Light Forest Biome before doing any analysis.
  • There seems (?) to be a threeway synergy between Warriors, Woodcutters and Crafters in this biome. (Ad hoc explanation: Woodcutters cut down trees, Crafters make things from the wood, Warriors stop the Elves from retaliating).
  • More specifically, it seems like the two sources of income are wood-based goods (per the point above) and ore-based goods (Miners / Smiths / maybe Crafters here too). This is important because ore-based goods need Miners, and we can't have >4 Miners without exposing ourselves to more danger than I'm comfortable with. 
  • Therefore, my strategy is to get as much as possible from ore-based goods (which seem more profitable in general, and don't suffer from diminishing returns w.r.t mandwarfpower), but fill in the gaps with wood-based goods.

The most important detail:

I have decided to call my fort Treeslaughtered.

Comment by abstractapplic on Ars D&D.Sci: Mysteries of Mana Evaluation & Ruleset · 2022-07-19T11:16:27.511Z · LW · GW

I liked this one a lot. In particular, I appreciate that it defied my expectations of a winning strategy: i.e., I couldn't get an optimal or leaderboard-topping solution with the "throw a GBT at the problem, then iterate over possible inputs" approach which won the last two games like this.

I think the Dark mana thing was a good sub-puzzle, and the fact that it was so soluble is a point in favor of that. It seemed a little unfair that it wasn't directly useful in getting a good answer, but on reflection I consider that unfairness to be a valuable lesson about the real world: the most interesting and cleanly-solvable subproblem isn't necessarily going to help you solve the main problem.

Comment by abstractapplic on My Opportunity Costs · 2022-07-11T11:47:54.225Z · LW · GW

You make a valid point, but . . .

basic encryption

The 'basic encryption' you have in minds is a Computer Thing. To the journalists in question, it was a New Computer Thing. If you're a Computer Person, you're probably underestimating the reticence associated with attempting New Computer Things when you're not a Computer Person.

much easier to use

I think that's false, albeit on the merest technicalities. The OTP system I have in mind is awkward and time-consuming ( . . . and probably inferior to Tor for Wikileaks' use case), but in terms of easiness it's something you could (probably) (eventually) teach the average (sufficiently motivated) tweenager.

Comment by abstractapplic on Ars D&D.sci: Mysteries of Mana · 2022-07-10T11:36:06.894Z · LW · GW

I took an ML-based approach which gave me radically different answers; the machine seems to think that

Matching currently-strong mana types is much more important than countering your opponent's choices.

As such, my new best guess is 

Fireball, Rays, Vambrace

Which should

give my master roughly 2:1 odds in favor.

Comment by abstractapplic on Ars D&D.sci: Mysteries of Mana · 2022-07-10T02:28:42.828Z · LW · GW

Oh, also:

I deduced the existence of Darkness Mana, determined that it almost certainly has a value in the 16-18 range, and then . . . couldn't figure out any clever way to use that information when strategizing. I suspect I'm missing something here.

Comment by abstractapplic on Ars D&D.sci: Mysteries of Mana · 2022-07-10T02:25:22.066Z · LW · GW

My provisional answer is:

Fireball, Levee, Hammer

This is supported by the reasoning that:

Levee (Fire/Earth) does a passably mediocre job protecting against Missiles (Earth/Water) and Fireball (Air/Fire); Fireball (Air/Fire) and Hammer (Light/Air) can both sneak past Solar (Fire/Light) by sharing an element.

And more prosaically by the fact that:

When I filtered the dataset to have Wizard A with the opponent's spell list, the spells which most raised Wizard B's winrate were those three.

However:

I've had a hard time figuring out how to weight "counter the opponent's element choices!" vs "go with what has the most ambient mana!" vs "go with what blocks the opponent's highest-mana attacks!". It's entirely possible that I should replace Hammer with Missiles, Rays or Vambrace; I hope to look into these possibilities later on.

Additionally:

The opponent picked a pretty good set of spells for the conditions in play; as such, I'm seriously questioning whether I can get my master even a >66% winrate.

Comment by abstractapplic on Contest: An Alien Message · 2022-06-27T15:11:38.194Z · LW · GW

Misc. notes:

  • As we've all discovered, the data is most productively viewed as a sequence of 2095 8-byte blocks.
  • The eightth byte in each block takes the values 64, 63, 192, and 191.  64 and 192 are much less common than 63 and 191.
  • The seventh byte takes a value between 0 and 16 for 64/192 rows, weighted to be more common at the 0 end of the scale. For 63/191 rows, it takes a value between ??? and 256, strongly weighted to be more common at the 256 end of the scale (the lowest is 97 but there's nothing special about that number so the generator probably has the capacity to go lower and just never exercised it during the generation process).
  • I agree with gjm that at least the first six bytes should probably be read as little-endian fractions. The first ten lines are variations on "x/5, expressed in little-endian hexadecimal, with the last digit rounded": notice how there's an 'a' in the early 999 row and a 'd' in the early ccc row, but no equivalent for the early 000 or 666 rows. And applying this reasoning to the rest of the data gets a lot of very neat fractions . . . for the first 80 rows or so, after which things rapidly degenerate.
  • I can't find any more correlations between these features, at least on a per-row basis. Between rows . . . there are a lot of 'pairs' of rows in which there's a 63-row followed by an otherwise-identical 191-row (or 191 by 63, or 64 by 192, or 192 by 64). These pairs are usually but not invariably seperated from the next pair by at least one unpaired row.
Comment by abstractapplic on Contest: An Alien Message · 2022-06-27T11:13:22.757Z · LW · GW

If (like me) you're having a hard time reading the .bin format, here's a plaintext version of it in hexadecimal.

Comment by abstractapplic on D&D.Sci June 2022 Evaluation and Ruleset · 2022-06-13T13:00:01.171Z · LW · GW

Confirmed and corrected; thank you again.

Comment by abstractapplic on D&D.Sci June 2022 Evaluation and Ruleset · 2022-06-13T10:46:09.961Z · LW · GW

Yes, good catch, fixed now.

Comment by abstractapplic on D&D.Sci June 2022: A Goddess Tried To Reincarnate Me Into A Fantasy World, But I Insisted On Using Data Science To Select An Optimal Combination Of Cheat Skills! · 2022-06-13T10:22:48.259Z · LW · GW

I would give you more time, but

you've already reached an optimal answer.

(Also, you can always just refuse to read the ruleset until you're done with the data.)

Comment by abstractapplic on D&D.Sci June 2022: A Goddess Tried To Reincarnate Me Into A Fantasy World, But I Insisted On Using Data Science To Select An Optimal Combination Of Cheat Skills! · 2022-06-05T09:35:43.779Z · LW · GW

As DM, I can confirm that skills provided with the help of the Chaos Deity or Eldritch Abomination are identical to those provided by the goddess alone.

Comment by abstractapplic on D&D.Sci June 2022: A Goddess Tried To Reincarnate Me Into A Fantasy World, But I Insisted On Using Data Science To Select An Optimal Combination Of Cheat Skills! · 2022-06-04T11:19:57.738Z · LW · GW

Nope; cheats are commutative.