Troll Wizard is random "good" character selection?

[rebthor]

I started a game yesterday and asked the RNG for a random good and got TrWz which doesn't seem like a random good to me as *all* the Wz skills have significant malus for a Tr and things that will keep you alive like some unarmed have to be trained from scratch.

I asked this on IRC but I'd like to raise it again, namely how is this random "good?". The answer I got why it's a recommended combo was "It's a Troll," which I can kind of understand but on a slow leveling race, having slow leveling skills seems like an automatic death sentence. I'm no expert player but I at least knew enough to turn off everything except unarmed to get some survivability. Now I can see TrTm due to the overlap in unarmed fighting and TrEE where the apts on a Troll aren't terrible as being "random good."

I readily concede that Wz is generally a really good background and Tr are generally a really good (starting) race but just because steak is good and ice cream is good doesn't mean that steak bits in ice cream is good. You may as well say that Tr** is recommended or Sp** or **Be but I know that I couldn't in good faith recommend a DEBe or SpGl to someone who wasn't looking for a challenge.

[Dustbin]

Wz is really strong early, Tr is really strong early.

Getting meph/conj flame castable doesn't take huge investment, and hunger costs don't mean anything to you.

[rebthor]

Recommending every Tr except CK and Wn seems reasonable to me. But yeah, it's ridiculous that TrWz is recommended and TrIE/TrTm/TrWr aren't.

Also, unrelated, but I have no idea why you would expect SpGl to be bad.

That's precisely my point, SpGl are at least as viable as TrWz if not more. Yet it's not "recommended." The underlying question is what is "random good" supposed to give you? TrWz may be powerful but it's going against type and it's not something that someone who has played a decent amount of crawl is going to know anything to do with especially since (and feel free to correct me if you disagree) the optimal play is to not train any of the starting skills and instead get a decent base of unarmed up.

[crate]

You probably want to get meph castable with TrWz, since meph is pretty good when you kill stuff with actions that cost 0 MP (unarmed). That will take some skill investment.

I think keeping it recommended is fine, it's a good combo to hammer home the point that a background is just a starting point. I agree with minmay that TrIE/TrWr at least should be recommended also (and TrSt is probably pretty good because Tr with evap is even better than Tr with meph, though the rest of the starting book isn't very useful for trolls).

[rebthor]

I'd also say that TrTm and TrEE should be in there based on past experience playing local. TrTm since they start with a good enhancer to UC and in theory if you spend the XP, statue form allegedly makes them even stronger. While I didn't get that far, evap is cheaper than meph in the old book and spider form is useful to run away or avoid hard hitters. Only one spell in the starting book (both old and new) isn't that useful, namely Blade Hands. It's good, but I don't know if the XP needed is worth it the same way it is for other Tm.
On the EE side, you get stoneskin which is very good for Tr since they're limited on armor. You also get passwall which is situationally useful.

[mikee]

There is no real science behind the recommended combos, and they have always been at least a little bit arbitrary.

Originally, most combos were locked. For instance, if you wanted to play a mummy, you could only play a fighter, necromancer, or death knight. I wasn't using IRC when the rest were unlocked, but I was told that someone (Jeff, I think) convinced the dev team to unlock them by compiling a list of combos he thought were good. The rest would be available to play but grayed out.

A few months ago, we (mostly elliptic and marvinpa) revised the list by haphazardly clicking around a google doc. I don't think recommended combos are meant to be perfect; there is always some subjectivity in what is 'good' or 'bad' to play.

[galehar]

The question is: should we recommend strong combos (easy to win) or easy combos (simple to play)? For example, DDNe is a very strong combo, but not an easy one I'd say (and TrWz is also, although not as much).
The original purpose of recommendations was to guide newbies through the myriad of possible choices. Easy combos were recommended. But when more experienced players ask for a viable combo, they expect to get a strong one. So as time went by, I think the focus drifted toward strong combos.
Here is an idea: we remove the viable choice and random always select among the recommended choices. But you can select your "player skill": beginner, good, expert. In beginner mode, easy combos are recommended, in good, strong combos are and in expert, they are all recommended (so this is were the old random choice is).
Unfortunately, I have the feeling that it's too much work to code and maintain for what it's worth.

[XuaXua]

The question is: should we recommend strong combos (easy to win) or easy combos (simple to play)? For example, DDNe is a very strong combo, but not an easy one I'd say (and TrWz is also, although not as much).
The original purpose of recommendations was to guide newbies through the myriad of possible choices. Easy combos were recommended. But when more experienced players ask for a viable combo, they expect to get a strong one. So as time went by, I think the focus drifted toward strong combos.
Here is an idea: we remove the viable choice and random always select among the recommended choices. But you can select your "player skill": beginner, good, expert. In beginner mode, easy combos are recommended, in good, strong combos are and in expert, they are all recommended (so this is were the old random choice is).
Unfortunately, I have the feeling that it's too much work to code and maintain for what it's worth.

I think you should adopt the skill menu coloring.

Blacked out if N/A, gray if available to be selected (to distinguish further from the N/A choices) and white as recommendations.

For a long time, I did not realize you could select the dark gray choices.

[rebthor]

The question is: should we recommend strong combos (easy to win) or easy combos (simple to play)? For example, DDNe is a very strong combo, but not an easy one I'd say (and TrWz is also, although not as much).
The original purpose of recommendations was to guide newbies through the myriad of possible choices. Easy combos were recommended. But when more experienced players ask for a viable combo, they expect to get a strong one. So as time went by, I think the focus drifted toward strong combos.
Here is an idea: we remove the viable choice and random always select among the recommended choices. But you can select your "player skill": beginner, good, expert. In beginner mode, easy combos are recommended, in good, strong combos are and in expert, they are all recommended (so this is were the old random choice is).
Unfortunately, I have the feeling that it's too much work to code and maintain for what it's worth.

I think you should adopt the skill menu coloring.

Blacked out if N/A, gray if available to be selected (to distinguish further from the N/A choices) and white as recommendations.

For a long time, I did not realize you could select the dark gray choices.

I don't have a problem with indicating which are "beginner" combos and which are not, but it might be better to have them as "bright" white and some as just white to work around the issue you encountered. Gray might already be too much and make people think they couldn't choose it.

[XuaXua]

The question is: should we recommend strong combos (easy to win) or easy combos (simple to play)? For example, DDNe is a very strong combo, but not an easy one I'd say (and TrWz is also, although not as much).

DDNe would be one of the first combos I'd recommend to a new player if they just wanted to win...in fact, the only combo I'd consider simpler to play is DDEE. I don't think the difference is very large, except for those few combos that have easy/streakable early games but are relatively hard later, such as DEBe or many Og/Dg.

Definitely have to add that they need to abuse that wand because they won't heal normally.

[TehDruid]

Agree about the hints mode; It really is where all the information a newbie would need to make a choice on a character combo should be. And it succeeds on doing so very well. Honestly, with the work that has been put into the tutorials and the hints mode I'd say you're already giving away too much to the newbie and that should be appreciated. Other roguelikes leave you totally clueless as to what you should be doing.

[XuaXua]

Do note that this is one of the roguelikes that has been the subject of student usability reviews.

[tormodpwns]

I'd also say that TrTm and TrEE should be in there based on past experience playing local. TrTm since they start with a good enhancer to UC and in theory if you spend the XP, statue form allegedly makes them even stronger. While I didn't get that far, evap is cheaper than meph in the old book and spider form is useful to run away or avoid hard hitters. Only one spell in the starting book (both old and new) isn't that useful, namely Blade Hands. It's good, but I don't know if the XP needed is worth it the same way it is for other Tm.
On the EE side, you get stoneskin which is very good for Tr since they're limited on armor. You also get passwall which is situationally useful.

Blade Hands is still effective on Trolls. They got high strength, which translates to extra UC.

[rebthor]

I'd also say that TrTm and TrEE should be in there based on past experience playing local. TrTm since they start with a good enhancer to UC and in theory if you spend the XP, statue form allegedly makes them even stronger. While I didn't get that far, evap is cheaper than meph in the old book and spider form is useful to run away or avoid hard hitters. Only one spell in the starting book (both old and new) isn't that useful, namely Blade Hands. It's good, but I don't know if the XP needed is worth it the same way it is for other Tm.
On the EE side, you get stoneskin which is very good for Tr since they're limited on armor. You also get passwall which is situationally useful.

Blade Hands is still effective on Trolls. They got high strength, which translates to extra UC.

As I said, "It's good, but I don't know if the XP needed is worth it the same way it is for other Tm." Obviously if you're going to get statue form, blade hands will be worth the XP investment. For a troll, since they keep the bonus claw damage in statue form, it's not as useful as "regular" blade hands is.

Statue Form gives 6+(str/3) + 9 base damage from claws + unarmed skill * 150%. While that 150% is effective cancelled out by action speed but it will increase the ability to break through high AC. Not that the knowledge bot for unarmed combat says that statue is 3+(st/3) damage. I'm not sure which is correct and don't have time now to code dive.
Blade hands gives 12+(STR+DEX)/4+unarmed skill.

[TehDruid]

Actually TrWz is the best package for INT for Trolls (and for any race, of course). You start out with 14/14/8 which is great if you want to dabble with spells.