Discussion What is the purpose of This Entire Website?

[Moyashi]

Furthermore, as someone that does some professional writing, my experience is that the hardest part is getting words on the page. After that, editing it and making tweaks is the easy part. The cases for which this doesn't apply would be people with weaker writing skills and people for whom English is not their first language...but then, wouldn't the availability of AI as a tool broaden our pool of potential contributors to such people?

Editing and tweaking writing is never “easy” either. Especially if it’s a pretty large piece of writing, It can be a pretty involved process that potentially goes through multiple phases. This gets more complicated if you’ve got peers involved reviewing your work to help iron things out, as you may have differing ideas on what needs changing or if something needs to be done one way or another. It’s frustrating sometimes to get stumped during the process but when that happens you reach out to people for advice. AI ain’t necessary nor does it do it better at all (not even close).

Regarding “cases like people with weaker writing skills and people for whom English is not their first language”, no to both. AI is not going to help someone with weaker writing skills write a good analysis, just like it doesn’t help a beginner artist make better art. There’s fundamental skills they have to learn in both fields that AI is never going to give them, and if anything relying on it will only lead to them leaning on it as a crutch and prevents them from improving. When it comes to language and translation, it’s not even good at that. Notoriously so (there’s a reason why a certain twitter AI translation tool is widely laughed at for how poor it is).

The point I'm interested in discussing here is whether analysis-writing is a fun and engaging process, which I argue it isn't. If you disagree, share your perspective!

Even if you think the current process is not fun or engaging, which in a vacuum is fair criticism by itself, suggesting the use of AI is just entirely undermines it because the use of AI itself in anything creative (writing, music, art, etc) is not fun in the least (on top of the ethical problems around it).

The analysis writing process can always been improved, but you’re not improving it through the use of AI.

 

[Finchinator]

I do not think we need an entire Policy Review thread on the ethics of using AI for Smogon. I most certainly do not think we need to hijack a previously existing thread to do so either.

I feel any further dialogue on the topic is doing a disservice to our network of skillful and driven contributors to On Site content, too.

 

[chimp]

Im a former OU C&C mod, I wrote quite a few analyses + QC’d quite a few more. Believe me when I say that using a LLM would barely be any help at all. The amount of work needed to feed the AI all the relevant info then sift through it for inaccuracies / writing quality would make switching to it permanently at best a lateral move. And considering the ethical concerns… it would be a wholly pointless switch. I mean there is a reason why the Translation team doesn’t use Google Translate.

I agree that we should allow for more creativity in the analyses themselves, and the process could be streamlined, made less tedious. But I also think to a certain extent the writing for analyses are standardized because… they have to be. It’s like wikipedia. This is formal, expository writing. It is sort of supposed to be a little dry?

If writing that kind of writing is unappealing for people then they could write for the Flying Press instead.

 

[Sabelette]

I don’t disagree that the templated and surface level writing of analyses is outdated and structurally doesn't look much different from asking an LLM to write an analysis but let’s not kid ourselves that LLMs can handle even the most basic shit.

Check this out

I highlighted stuff that’s wrong and added comments of my own
_________________________________________________________________
Write me a basic explanation of Tauros movesets and how to use it in RBY OU

ChatGPT said:​

Tauros in RBY OU (Red/Blue/Yellow - Original Generation)​

Tauros is one of the most iconic and powerful Pokémon in the RBY OU (OverUsed) tier, primarily due to its excellent stats, versatility, and access to powerful moves. Here's a breakdown of its movesets and how to use it effectively in RBY:

Key Stats:​

• HP: 75
• Attack: 100
• Defense: 95
• Special: 40 (off to a great start as it hallucinates that Tauros's SpA is its Gen 1 Special)
• Speed: 110

With 110 Speed, Tauros is faster than most threats in RBY, allowing it to outspeed and hit hard with powerful physical moves. Its 100 Attack is also very solid, making it a great offensive presence.

---

Common Movesets for Tauros in RBY OU:​

1. Standard Physical Attacker
   Moveset:
   • Body Slam / Takedown (lol???)
   • Hyper Beam
   • Earthquake
   • Blizzard / Explosion (lmao?)
   • Explanation:
   • Body Slam is preferred by many because of its chance to paralyze the opponent, adding utility in addition to raw damage.
   • Takedown is an alternative to Body Slam if you want more damage, but it comes with recoil. (yep lemme trade 15 accuracy and 30% para chance for... 5 more base power. Forget Double-Edge, this is where its at)
   • Hyper Beam is a powerful finishing move that can often KO an opponent after Tauros has weakened them with other attacks. (even the most dogshit generic analyses aren't this bad at saying anything of substance)
   • Earthquake provides coverage against Fire, Rock, and Electric types, which are common in RBY OU. (RBY OU: one Rock type, zero Fire types, 2 Electrics and one is also Flying-type)
   • Blizzard is often used to deal with Dragonite and other flying or Grass types. Alternatively, Explosion is an option if you want to go for a sacrificial play to take down a key opponent. (Dnite isn't common and Blizz only does slightly more to Zapdos/Eggy than Body Slam; Blizzard is more for Rhydon than any of those)
2. Choice Band (if possible in RBY) (LMAO????)
   While the Choice Band item wasn’t available in RBY (it was introduced in later generations), Tauros' effectiveness in RBY as a powerful physical attacker can be compared to how modern Choice Band sets work. The idea is to choose one offensive move and rely on its power. In RBY, Tauros has access to high-powered moves like Hyper Beam and Earthquake to hit hard. (How does anyone think AI is a reliable source of info on anything)
   • Agility (How many moves will this AI invent despite being able to pull directly from an actual existing analysis already on this site)
   • Body Slam
   • Hyper Beam
   • Earthquake
   • Explanation:This set aims to use Agility to double Tauros' Speed, allowing it to outspeed more threats. After Agility, Tauros becomes a late-game cleaner that can hit almost everything hard with Body Slam and Hyper Beam. However, be cautious as it requires a setup turn, during which Tauros is vulnerable.

---

How to Use Tauros Effectively in RBY OU:​

1. Lead or Mid-game Pressure: Tauros can be used as a lead or a mid-game threat to apply consistent pressure. With its high Speed and offensive presence, Tauros can hit hard and force switches. A strong Body Slam or Takedown can break through many of the bulky Pokémon in RBY. (More dogshit quality writing and dubious info)
2. Cleaning Up Late Game: After a few of the opponent’s Pokémon are weakened, Tauros can clean up with a Hyper Beam or Earthquake. Its speed allows it to outspeed and KO key threats, often swinging the momentum of a game in the later stages.
3. Status Immunity: Tauros is a great counter to paralysis-prone Pokémon due to its Clear Body ability (though it doesn't exist in RBY; it's a nod to its future abilities). However, it's important to avoid being statused with Thunder Wave or Sleep because Tauros needs to remain mobile and deal damage quickly. (This is my favorite one. Tauros counters... Pokemon weak to paralysis. Because it has Clear Body. Which makes it immune to status. But also you have to avoid being statused.)
4. Handling Physical Walls: Tauros' Hyper Beam and Earthquake are great tools to break through physical walls. However, be cautious of Reflect-set-up walls that can tank Tauros' hits and retaliate.
5. Avoiding Special Attacks: Tauros' Specialstat is low, and thus, it is vulnerable to special attackers like Alakazam and Exeggutor. Always keep in mind the threats from fast Special attackers that can take advantage of Tauros' low Special defense. (70 Special or in later gens SpDef isn't even that bad but it's treating this as 40)

---

Weaknesses to Keep in Mind:​

• Special Attackers: As mentioned, Tauros doesn't have great Special Defense, so Pokémon like Alakazam, Jynx, and Exeggutorcan be problematic if they can outspeed and hit with powerful special moves.
• Status Effects: Tauros can be crippled by Thunder Wave, Sleep Powder, or Burn. Keeping it healthy and avoiding status is key to its success.
• Strong Physical Walls: While Tauros can hit hard with Hyper Beam or Earthquake, walls like Blissey (if played in RBY) or Chansey can shrug off Tauros' hits and heal. (Noted Physical Walls Blissey and Chansey. Notable RBY Pokemon, Blissey.)

---

Conclusion:​

Tauros is a versatile and fast physical attacker in RBY OU, capable of dishing out huge damage with its strong offensive movepool and high Speed. It can function as a powerful physical sweeper or an aggressive mid-game wallbreaker. Its Hyper Beam and Earthquake make it a threat to a wide range of Pokémon, but it does need support to handle faster special attackers and status effects.
When using Tauros, try to keep the pressure on with its speed and offensive moves while making sure you don't let it become a sitting duck for faster, special attackers.

Like, all of this was complete useless dogshit, it omitted a ton of important information, it's riddled with typos and stuff that would not be acceptable in GP, and even just reading that and pointing out issues took longer than it would've taken me to write an analysis with some actual context and utility beyond explaining "X move hits Y super effectively." And this is for a Pokemon that has 30 years of existing discussion around how to use it including an analysis on site already that this is absolutely reading as part of its data set. Anyone who thinks this is a useful tool and not just a forest-burning disinfo-spewing infinite stupidity-validator for people with worse critical thinking skills than my dog that has repeatedly run into glass doors needs to go back to grade school and try life over - it's about as good as Google on a good day and it never has a good day.

Anyway, if we want to talk about accessibility and the community, which is the only interesting part of this conversation to me, yes, we should talk about encouraging some more in-depth resources if people have the drive to make them. Sample teams without commentary and basic analyses are fine as a starting point and they help some people, though I agree many analyses are too rote and contextless as things stand. We could also do more in the form of replay analysis and extra context for sample teams and analyses for metagames as they age and stabilize, but the workload is quite large. I'm doing something of the sort by recording RBY set commentary on Youtube and focusing more on understanding team compositions, predicting last Pokemon after some are revealed, predicting sets, positioning, etc. rather than play-by-play commentary without any analysis, and I think we could use a lot more of that stuff, but it might be better suited for individual people in different communities making some low-to-medium production quality videos or forum posts rather than being ingrained as a standard for analyses and samples. It's just not practical to ask every single tier and community to adopt that kind of standard, even if I think every relatively-stable meta would benefit from having advanced resources.

I think passionate high-level work like this is the lifeblood of a community, and every little subgroup on Smogon lives or dies on the passions of a few people who work hard on developing, discussing and building out each metagame. I think we could reward and incentivize the people who want to do higher-level stuff than writing "X uses coverage move Y to hit meta staples Z and A" over and over and try to move analyses and samples toward increased context, while still acknowledging that that sort of super basic info has a use for lots of players that aren't nearly as invested as the people typing paragraphs about niche site policy. I'm not really sure exactly what the best way to do that is, because I think attaching badges to everything also attracts a lot of badge-grubbers that churn out work to get pixels on their profile or a custom avatar and clearly there needs to be some level of quality-gating if this stuff is going up on sample teams or the dex, but maybe others have better ideas on this than me.

 

[TheDuckChris]

Im backing off of LLMs cause the rest of this post is more interesting to me.
The single purpose of this site has always been to play Pokemon and make it into a better game than it is when fighting in game trainers.
I think we agree on this because the rest of this post seems to discuss "what does it mean to make a better game".
So I will look at this through a game design lens.

Resources - this is the tutorial or game guide. Most games these days have a wiki that supplements the tutorial and acts as a helper for players to figure out the right way to play, how to get past frustrating spots, and generally just an all-effective "help button". This encompasses all of C&C, sample teams, viability rankings, RMT, tutors, and metagame discussion threads. This is also the hook, the thing that convinces and allows new people to play the game

Tiering - this is the rules of the game. Playing by our rules makes the game a more complicated and deeper game than the cartridge. Multiple tiers allow for players to expand their gameplay into different venues. It broadens the flavour appeal, allowing for your favourite pokemon to be used in the game regardless of how bad they are (for most). They can be restrictive and even vague, but playing by the rules generally makes the game better than not.

Tournaments - These are extra game content. When you've gotten tired of the base game (ladder) you can switch up your gameplay to a new format. It's the same game, with a new points system and new rewards.

Social stuff - every game has an online community. With smogon, the forum is the community as well as the game. This is the MMO aspect of the game, allowing for personal social interaction that both enriches the gameplay and is enriched by the gameplay in turn, making for more exciting and interesting social interaction than your standard chat room. This also extends to all sorts of forum based social activity, including badge hunting, profile customization, PR debating, tournament trash talking, and of course the actual social sections.

Moderation

Every game needs devs and also moderators. Pretty self explanatory, we have those. Unlike most games the moderators and developers are unpaid volunteers and any internal team problems can become site-wide drama on a semi-regular basis.

---
There's plenty more to discuss re: the health and status of these parts of the site, but I'll stop this post here as I think it's a good starting point and it's hard to read a lot of text. Do I have it right? What parts of the site are missing or don't fulfill their roles properly?

 

[fiish]

answering questions

A game is considered "solved" when you can consistently maximize your odds of winning with either an unchanging pure strategy (always bring X team) or unchanging mixed strategy (40% chance to bring X team, 60% chance to bring Y team). Is it desirable for a metagame to be solved? Why or why not?

If a metagame is solved that means that it's extremely simple and not really worth playing anyways. The only actual solved metagames are memes where you have <=6 real pokemon and very poor movepools. Actually solving any metagame that's more complex is probably impossible.

Are more unbalanced or uncompetitive metagames generally closer to being solved? Is this always the case, or are there cases where a very unbalanced metagame is unsolvable?

This is probably true because in a theoretical "fully unbalanced" metagame either one strategy is obviously dominant and you should always pick that, or the metagame is so matchup dependent that it becomes rock-paper-scissors (a solved game).

Consider the metagame NU + Arceus. Due to Arceus's extreme versatility, it's difficult to come up with a specific winning strategy. Can a more vague strategy such as "Limit opposing Arceus with X and Y measures, and build around any good Arceus set" be considered a solution? If this strategy is generally sound, does this limit the amount of ways in which you can think about the game? Does "use viable Pokemon, and knock out all 6 opposing mons" limit the amount of ways in which you can think about the game?

It's not really a solution because there probably exists an Arceus set that can beat (or at least put into a losing position) any given team. If that strategy does work, it probably does limit the amount of ways in which you can think about the game due to any plan needing to greatly consider a pokemon with very limited counterplay. I don't really understand the last question as knocking out your opponent's pokemon is how you win, which you'll always need to do, (unless you're in a losing position and playing for a tie) and what matters is how you accomplish that.

If you use the mixed strategy of taking one of the 17 sample teams at random, is BW OU a solved metagame?

Almost certainly not! For example, it might be that choosing from a subset of the sample teams at random actually leads to more wins, or choosing from the samples + some extra teams makes you win more. (This is basically what will happen when samples get updated.) And in terms of playing, using such a strategy will put you at an information disadvantage vs anyone who isn't using a sample team and who knows your sets.

Are the people playing whichever hax-removal mod is currently in fashion playing an objectively better game than we are? Ok, I hear you, you say those things can't be objective. Are they playing a subjectively better game than we are?

Probably, but there's a lot less players.

Let's introduce another word, "complexity", and define it as "the amount of computation it takes to solve a game". So if a metagame is easily solved, it isn't complex, and vice versa. If complexity is desirable, is more complexity more desirable? Why or why not? Is there any limit? If so, why is it there?

If more complexity was always more desirable, then no one would play chess and everyone would play some variant with 20 more pieces. At a certain point a game becomes so complex that it's impossible to tell which moves are good and you might as well be making moves at random. I don't think pokemon can become this complex, though there still might be a format that's too complex for anyone to want to play.

Can an in-battle position be more complex for one player than for the other?

Yesterday I thought the answer was no but I think that if one player has a very rng-based move, like zap cannon, and the other player doesn't know they have it, there's a lot more to consider for the player with the move than the player without.

Can a battle become more complex over time?

Yes. If you have a game that's won by hitting hydro pump, and you miss, then the game has become more complex. On the previous turn, 80% of the time the game ended in a guaranteed win (very simple), if you miss, then 0% of the time the game ends in a guaranteed win and you have to "start playing" again.