Golden Sun 2 Optimisation Ideas

						Golden Sun 2 Proposed Route Improvements

Some scattered thoughts for the next person who wants to take up the mantle of routing this beast.

	RNG Manipulation

It's clear that RNG manipulation to run from all encounters up to King Scorpion is possible and desirable, so that's a no brainer. Things are way less clear once you enter Kibombo, and here's why.

Problem 1: Experience

Assuming RNG manipulation is possible, are you aiming to run from every encounter or are you taking fights? Both of these are problematic for different reasons.

If you run from every encounter then you're going to be severely underleveled. This wasn't such an issue in GS1 because there was frequent unskippable boss fights and Djinn to collect. This isn't the case in GS2, after King Scorpion the next boss is Briggs ... which happens AFTER Tundaria tower (Aqua Hydra is skipped) and after King Scorpion, there is 1 fought Djinn in Gambomba (none before that) and 2 more Djinn before Aqua Rock. That means our party is even more underleveled than our GS1 counterparts at a similar point in the game. The first real place that this would manifest as a problem would be Aqua Rock as the exterior is an RNG hell room. There's quite a lot of time spent here so realistically you'd be forced to save crawl since the mobs here are lethal even if you have been taking fights. The interior is slightly more manageable, but the flashy lights that are generated by using douse on a tear stone turn rooms into RNG hell rooms which greatly limited the strength of RNG manip in the interior.

Tundaria tower would in theory be doable, but RNG methods would be very difficult (as anyone who's attempted to manipulate Venus 1 will tell you, low level party RNG methods for high level mob fights are very difficult to do). Let's say you've survived to now minimising encounters.. there's no way that you would be able to do a 0-light serpent strat which means you're going to lose 7 minutes to a route which does take encounters. Meaning that the time save from taking no fights would need to vastly exceed this (Poseidon would also be a tough fight... but in theory he'd be doable even if it took a little longer. Moapa falls under the same category - doable but slow). For what it's worth, Moapa is essentially the final boss for this method because Doom Dragon can be done with Isaac's level 28 mob.

Okay, now that we've gone over the issues with 'no fights' let's look at the issues with taking fights. The first problem is level up management. When you level up, the GRN seed you are on determines the stat bonus you get. For the time being let's not worry about what the level up actually is, since that can be addressed later. Since we're routing out 6 GRN seeds for each segment of the game each of those seeds is going to produce a different total exp per segment (unless you keep the fights consistent across all the seeds which seems really slow). The problem with this is that the complexity for the total experienced earned grows exponentially - 2 segments = 36 different EXP combinations, 3 segments = 216 different combinations - you get the drift.

The good news is that with a little care this problem can be overcome but the person routing it needs to be aware of this. The idea is to create a lower bound for your possible total experience and an upper bound for your total experience. For each seed you make sure that the max experience and the min experience files level up during the same fight, and then ensure this happens for each of the 6 seeds.

So after King Scorpion (or whenever you decide to take fights), you would route out each of the 6 seeds for the next segment and ensure that the number of level ups is the same (i.e. all level up 0 times, once, twice, ...). Then take the highest experience seed and route out the next segment of the game (ensuring that the same number of level ups happen across the seeds). Now take the lowest experience seed and use the route developed for the highest experience file and ensure that any level ups occur during the same fight as the highest experience seed. Finally, take the lowest experience from the lowest experience file and the highest experience from the highest experience file and route out the next segment. My advice would be to keep the difference between the high and low experience files less than the average experience from a mob in the area you are currently in -- any higher than that and you risk running into some serious problems.

Problem 1a: Agility

Closely tied with the experience problem is the problem of agility and meeting the agility threshold for Serpent. I would advise that each level up is manipulated to keep the agility constant (as I've done up to King Scorpion). This way you can be sure of what the party agility is by Serpent which makes the manipulation easier. That said, you could also min/max it like I proposed above.

Problem 1b: When to start taking fights

It's entirely reasonable to only start taking fights after Gabomba. Why? Because the mobs after the ship are much weaker than those in Gabomba BUT they give approximately the same experience. This is the games experience catchup mechanism in case people are underleveled. That said, I don't think you'll earn enough experience to survive Aqua Rock without starting to take fights easlier.

Problem 2: Lemuria Fountain

In theory, you need only 3 lucky medals to get every thing you need (Eclipse and two Wild Coats). I believe my current notes have the count down to 4 or 5. Each coin toss takes up a considerable time so perhaps someone could look into reducing the count to three. Obviously the issue is that with 3 coins how do you tell what seed you are on? After the hard reset you leave the fountain and watch the movement of the Lemurians as their movement is dictated by GRN. Once you have sufficient information (normally 2-3 movements) you will know what seed you are on and manipulate accordingly.

Problem 3: Double Aka Manah Mobs

The only reason you would consider Isaac's mob over Felix's mob is that Felix's mob can't clear a double Aka Manah mob in 1 turn (it can kill one of the two and severely damage the other, but Jenna's attack (with Phaeton blade) isn't high enough to take out the second one). It's possible that Isaac's mob could clear both since Garet naturally has a higher attack and would be at a higher level so may have access to Annihilation (Death Leap is resisted, but still deals the greatest damage). So perhaps swapping Garet in over Jenna is a real option (but Jenna is so cool Y_Y).

Here's why you wouldn't swap in all of Isaac's mob over Felix's.
- 	Isaac drains his PP faster than Felix does because he has access to Potent Cure (10 PP) while Felix settles for Cure Well
	(7 PP). It's actually a substantial difference.
- 	Isaac's mob does not clear Magma Rock faster than Felix's mob. The access to better psynergy still results in the same time to
	clear mobs. In fact, in conjunction with the previous fact Isaac's mob was taking up to 2 minutes longer to clear the rock.
- 	Isaac's mob would need to be geared up, increasing the time of the post-reunion menu by a small bit.
- 	Aside from Aka Manah's, there's no mob in Mars Lighthouse that has the potential to be cleared faster by Isaac
- 	Using Mia over Piers makes a lot of sense on paper. By this point in the game Piers is primarily a caster and Mia does this
	better because she is faster and has access to better psynergy. When I tested this theory there was no substantial difference
	between Mia/Piers in the clear time of Magma Rock, i.e. the RNG of the number of mob encounters played a greater role in
	determining the time to clear magma rock than the choice of party member. This comes back to the fact that even better psynergy
	doesn't change the number of turns it takes to clear mobs. With that said, there are some mobs that Piers won't outspeed that
	Mia does, so it may be marginally faster on average to include Mia but much more testing needs to be done.
-	Ivan shouldn't be substituted for Sheba. All of the previous arguments apply here, except that Sheba outspeed everything
	anyway.	Ivan's position in the party list is important for the last third of the game anyway, as it's highly convenient to
	have him last.	That means there's no good reason to sub out Sheba. Ivan for Piers is worse because of the Djinn imbalance
	that creates.

Problem 4: Items

I think that the items obtained can be greatly optimised and possibly more sleep bomb strats introduced for various djinn fights. Here are some examples of what I mean.
- 	Is it really necessary to buy the Adepts Clothes at Naribwe? They certainly are nice to have for the PP bonus as well as the
	extra survivability. But if fights are being RN manipulated then perhaps it isn't strictly necessary?
-	Dark Dawn showed that Armor is pretty useless outside of providing agility bonuses (generally speaking). Perhaps the Serpent
	can be manipulated so that we can win the fight without needing to make a shopping trip stop in Champa.
- 	We currently manipulate the Staff of Anubis as a drop in Gabomba for the substantial Attack upgrade for Jenna/Sheba. Firstly,
	does this even save time? I presumed it did, but perhaps it doesn't since the mist sabre is acquired reasonably early on in
	Aqua Rock?
- 	Would manipulating for the Tartarus Axe in Tundaria save time (especially since we skip the lightning sword). My answer to this
	is probably not. My logic is that Piers starts to drop off the compared to Jenna's attack once Jenna goes into Brute class.
	This means that weapon upgrades for Piers aren't really that useful. Since the Cloud Brand and the Tartarus Axe are comparable
	in terms of strength, it seems wiser to just get the Cloud Brand and skip the Tartarus axe (giving Thanatos Mace to Piers).
- 	Clotho's Distaff can be ruled out as a weapon upgrade for Jenna since I've confirmed that it cannot kill the double Aka Manah
	mob, and therefore offers no faster mob clear time. That said, perhaps Rune Blade (from Magma Rock) and Garet strats could?
-	Currently we get the two easiest sleep bombs to get in the run - from Daila Inn and Madra Inn. A lesser known fact is that
	sleep bombs are dropped by Aqua Jelly's fought in the ship. Perhaps a method could be worked out to get the drop so we can use
	three sleep bombs? Generally speaking, it takes longer to manipulate for the sleep bomb that it does to get the bomb from the
	inns.
- 	Psynergy items. Unfortunately, these aren't really useful. Since there's such a large gap in the run between King Scorpion and
	Briggs there's really no opportunity for us to use them. Moreover, they need to be manipulated from fights as opposed to picked
	up from towns so would actually take a while to get.
- 	Do we actually need the sol blade? It actually takes a while to get because of the annoying puzzle. Worth checking at least.

Problem 5: Mars Lighthouse 2

The mars section of Mars Lighthouse 2 takes so god damn long. The pipe puzzle room could be the longest room in the entire game because of the ridiculous encounter rate there AND the fact you can get double Aka Manahs. This could be a place where manipulating for no encounters is a desirable thing (despite the fact there is an annoying use of Move in the middle of the puzzle which *could* make life very difficult).

Problem 6: Doom Dragon

Surely the DD fight can be done in a better way. Unfortunately my current version is the best I can do. I know someone like Werster could come up with an insane strategy to beat DD. Nevertheless, it should be something that anyone serious about running the game should look into.

Problem 7: Djinn

I think every Djinn we get is more or less essential. I can easily justify why each Djinn is obtained (much like GS1 any%) but there are still questions about whether or not this is optimal or not. I'm pretty sure cutting any further Djinn would be slower (except maybe one Venus djinn from Piers... probably the one from Prox?) but it might be that obtaining a few more djinn speeds things up. For instance, it could transpire that after Mars 1 teleporting to Apoji Islands and grabbing Haze opens up more Eclipse use which speeds things up over all. Similarly, maybe obtaining the Jupiter Djinn in SW Atteka could serve the same purpose? It's worth pointing out that I really doubt that this is the case, but it's still something that I encourage people to look into.

---------------

Thats all for now, I may add more if I think of anything.