Squelch design considerations

[Derakon]

I promised fizzix I'd start a thread about how squelch ought to work, so, here we go!

The most important thing about squelch is that you should be able to handle at least 95% of cases through the 'k' command interface. No going to the knowledge menu, just selecting an option when you want to ignore a specific item. The knowledge menu has its uses, but as a general rule when you're squelching something you're saying "I want to ignore this item and others like it for the future" -- going to the knowledge menu to do that requires you to find the item again in a frankly massive list, and it's thus bad usability.

That out of the way, what we need to do is come up with sensible categorizations for what kinds of items the player wants to squelch, from very specific (just squelch this one item) to very broad (squelch items that bear only a passing similarity to this item).

---[b]Current Vanilla[/b---
(i.e. not v4)

Our categories here break down by quality and by item type. However, the category that has the most inherent complexity is also the one that we do the absolute worst job of handling: ego items. Vanilla's ego-item categories are "excellent with no high resists", "excellent with high resists", and "splendid". Ignoring that players don't necessarily know what a "high" resist is (protection from hallucination? ), there are massive gaps here between the program's capabilities and what the players want, and the "tiering" of these categories also causes problems:

* I don't care about Slay Orc, but I do still want to see acid brands (both excellent with no high resists)
* I don't care about Infravision (splendid), but I do still care about Seeing (excellent with high resists)
* This Dagger of Slay Evil is pretty useless, but a Bastard Sword of same would be helpful (same item type)

What we should be doing here is stealing NPP's squelch categories wholesale. When you squelch, say, that Dagger of Slay Evil, you get given the following options (as best I can remember, it having been some time since I played NPP):

* Squelch this item
* Squelch all Daggers of Slay Evil
* Squelch all non-artifact Daggers ("ego and below" in other words)
* Squelch all weapons of Slay Evil
* Squelch all non-artifact weapons

If you're still interested in Slay Evil, just not on Daggers, then you can choose the second or third option; if you don't care about Slay Evil at all, then you choose the fourth; if you've decided that your kit is really rock-solid then you choose the fifth. This does a good job of meeting the user's needs and avoiding arbitrary categorizations.

Behind the scenes, of course, we just have a bunch of categories that we squelch on, and if all of the categories for a specific item are marked as squelchable, then we squelch the item. Thus if you choose the second option, then the categories "Dagger" and "Slay Evil" are marked as squelchable.

---v4---

This gets a lot more complicated, because ego items, as specific collections of abilities, are no longer the only interesting non-artifact equipment the player can see. Magical items can get any number of collections of affixes, and what's worse, they can scale up semi-arbitrarily -- you can find a useful weapon that's useful solely because it has pluses out the wazoo (stacked the Sharpness affix a dozen times, say). The increased variety of potential items means that there are more "categories" of items that the player might possibly care about, which means they need more expressive capabilities on the squelch menu.

However, short of just having a gigantic grid of possible item flags and having the user toggle them on or off (as in the knowledge menu), there are a few options we can add:

* Squelch all weapons with less than X% balance / heft. Most characters care only about one class of items, rendering the "other side" less than useful.
* Squelch all items with this combination of flags (or some subset thereof). It's not unheard-of to find a weapon that only has e.g. the added-light-radius affix, which is pretty dull. So to speak.
* Squelch this flag/item type. Any item that is of a squelched item type and has only flags that are squelched, gets squelched. So for example, if you squelch all Daggers and the poison-brand affix, then a poison-branded spear will survive, as will a dagger with a cold-brand affix, but not a dagger with no affix, or a dagger with any non-poison-brand affix.

Honestly, v4's system is still evolving and it's hard, as a player, to say what kind of squelch system we need yet. So I admit this isn't so well-thought-out.

[fizzix]

Questions:

What would the options be for something like elvenkind. Imagine a +2 large leather shield of elvenkind [8 + 15] with +2 stealth and rNexus. What options would show up on the k menu? Your proposed menu would look like this:

• This item only
• All large leather elvenkind shields
• All non-artifact large leather shields
• All elvenkind shields
• All non-artifact shields

But you can imagine other options like

• All shields with rNexus (as special resist)
• All shields with stealth bonus of 2 or less
• All leather shields with less than +15 AC bonus

I think this question really gets at the heart of the trouble I've had in thought-designing a good squelch menu. There are lots of times where every blessed item is useless unless it has telepathy. I don't want to squelch all blessed items on sight (assuming something like mass-ID is here) then I could miss the one with telepathy, which is exactly the one I want. I don't see how we fit this into the basic categories.

Another similar example. I have a ring of strength +4. I want to squelch all other rings that are +4 or weaker. Can this be accomplished? What about more complicated flavored items, like amulets of weaponmastery with multiple pvals? Say, I have one that is (+3, +3) but I'd be happy with one that was (+3 +4) or even one that was (+2, +4). How do we handle this?

Lastly, what happens when you pick up a defender weapon and you want to squelch it, but you haven't ID'd it yet. What appears when you attempt to squelch it? Does the game tell you it's a defender?

[Derakon]

The important thing to realize about the squelch system is that it doesn't need to completely automate the loot-consideration process. Players would much rather have to manually deal with more items than they would potentially miss out on an item that they care about. If you're interested in Elvenkind armor only if it has rDisen or rNether, then just don't squelch Elvenkind armor. Sure you'll get a false positives, but honestly it's not that many items; Elvenkind is pretty rare to begin with.

Squelch-by-pval would be nice; IIRC NPP still doesn't have that (though I could be wrong). Of course multiple pvals make it a bit harder, but not impossible. Just set a threshold for each pval -- if none of them meet the threshold then the item is squelched. You can take advantage of the consistent ordering of pvals here (i.e. the first one on Trickery is always +DEX). Weaponmastery doesn't actually have multiple pvals, so let's use Trickery as an example:

*) Squelch all Amulets of Trickery with less than <+3> to DEX
*) Squelch all Amulets of Trickery with less than <+2> to Stealth
*) Squelch all Amulets of Trickery with less than <+99> to Searching, Infravision

What this says here is that the player doesn't care about the third pval, but any amulet that has at least +2 Stealth or at least +3 DEX will survive being squelched.

You can extend similar logic to hit/dam/AC bonuses/total AC/etc. but frankly I don't think it's worth the effort for Vanilla. It'd be more useful for v4, since you can get unboundedly high pluses on items thanks to affix stacking.

[buzzkill]

In order to round up the outliers, one could always include a secondary logical consideration such as...

Regardless of other squelch settings,
Never squelch <this item type> with <this property>.
Never squelch anything with <this property>.
Never squelch <this item type>.

Wouldn't that do the trick?

[fizzix]

@Derakon: Ok, I see how your original system would provide an improvement over current V behavior, even if it's not perfect. I don't know how to make a nice UI interface for the trickery situation. It's going to look pretty ugly.

Regardless, some UI improvements to squelch could be a nice target for 3.5 if someone wants to pick up on it.

@buzzkill: Would you have to go to the squelch menu to get that option? Or could it somehow be inserted into the 'k' interface?

[buzzkill]

@buzzkill: Would you have to go to the squelch menu to get that option? Or could it somehow be inserted into the 'k' interface?

The particular options I just laid out would probably be considered "advanced" and better suited to the options menu or wherever such advanced settings would be accessed.

I once envisioned a system where (y/n/s) would be the standard k dialog and then upon taking the staircase, so as not to disrupt the flow of the game, the player would be asked a set of logical (y/n) questions derived from the items he destroyed on that level, in order to fine tune/expand the squelch settings. Answering these question would be optional, but answering would result in better squelch.

[nppangband]

[QUOTE=Derakon;73340]

Squelch-by-pval would be nice; IIRC NPP still doesn't have that (though I could be wrong). Of course multiple pvals make it a bit harder, but not impossible. Just set a threshold for each pval -- if none of them meet the threshold then the item is squelched. You can take advantage of the consistent ordering of pvals here (i.e. the first one on Trickery is always +DEX). Weaponmastery doesn't actually have multiple pvals, so let's use Trickery as an example:

QUOTE]

Correct. NPP doesn't have that. You can squelch by specific object, or ego type (ie once you have found a shield of resistance, you have no need for shields of resist acid, fire, etc).

NPP also has pickup options by specific object that override the generic pickup options. For each object kind, you can set to squelch, always pickup (for somebody who has never_pickup on, but wants to always pick up potions of life), never pickup (for somebody who who picks up every object, but doesn't want to pick up, for example, stat gain potions after they are at 18/100, but wants to leave them on the floor and visible in case their stat gets drained).

One simple thing to add, NPP also has have high level weapons and armor in its own category, such as dragon armor and shields, blades of chaos, maces disruption, etc).

But you are correct that NPP doesn't yet have pval squelch. That's a great feature I need to steal.

[Patashu]

IMO as much as possible about squelching should be done for you. E.g. if potions of slowness are always bad, always squelch potions of slowness. If the game can determine you'd never want that helmet of infravision anymore since you have ESP in that slot, it should squelch automatically. Etc. The best choice is the one that you don't even have to make.

[Derakon]

IMO as much as possible about squelching should be done for you. E.g. if potions of slowness are always bad, always squelch potions of slowness. If the game can determine you'd never want that helmet of infravision anymore since you have ESP in that slot, it should squelch automatically. Etc. The best choice is the one that you don't even have to make.

That's fine as long as everyone agrees that the game makes the right decisions. But if there's a situation in which the game hides loot that a player wants because the game assumes they don't want it, that player will be peeved.

For example: mindless, warm-blooded monsters. I think giant salamanders qualify; if you had good infravision you could still see them even if telepathy didn't reveal them. It just so happens that all of the monsters that fall into that category are weak enough that the infravision vs. telepathy tradeoff isn't generally one that gets consideration, but the fact remains that you can't always unequivocally state that one option is better than another.

[Patashu]

That's fine as long as everyone agrees that the game makes the right decisions. But if there's a situation in which the game hides loot that a player wants because the game assumes they don't want it, that player will be peeved.

For example: mindless, warm-blooded monsters. I think giant salamanders qualify; if you had good infravision you could still see them even if telepathy didn't reveal them. It just so happens that all of the monsters that fall into that category are weak enough that the infravision vs. telepathy tradeoff isn't generally one that gets consideration, but the fact remains that you can't always unequivocally state that one option is better than another.

The default options for a game should be the right options for the majority of people who play the game, not the minority. ESP is better than infravision, so make ESP > infravision by default. Someone who knows they want infravision more than ESP also knows enough to dig into the options and change it to be what they want. In general, every A vs B choice that strikes you as 'you would only take option B under very, very weird circumstances' should be set to A by default.

[Derakon]

The default options for a game should be the right options for the majority of people who play the game, not the minority. ESP is better than infravision, so make ESP > infravision by default. Someone who knows they want infravision more than ESP also knows enough to dig into the options and change it to be what they want. In general, every A vs B choice that strikes you as 'you would only take option B under very, very weird circumstances' should be set to A by default.

But you're positing a false dichotomy. The options are A, B, or let the player decide. What I'm saying is that whenever A is not absolutely unequivocably better than B, we should leave it to the player. That includes not just weird edge cases like infravision vs. telepathy, but also concerns about weight and armor class and other such factors (what if you want to keep two sources of Free Action because you have an item back in the home that you plan to swap in next time you recall?).

It's fine if the game makes recommendations. It's not fine if the game assumes it knows better than the player does.

[Patashu]

Imagine if in the next version of Angband there were changelog notes along the lines of:

To help quell TMJ, we have decided to stop the spawn of less interesting thing as the game progresses:
-Starting at depth 35, weapons of slay orc will no longer be eligible to spawn.
-Starting at depth 30, helmets of infravision will no longer be eligible to spawn.
...

Etc.
For a given change like this, if no one complains about the change, then it seems equivalent that no one would complain about an equivalent on by default squelch rule (where said thing would start getting squelched when you had in that slot something that was a 'no brainer' to equip over it - e.g. fire brand or extra attacks + does more avg damage than the thing that tried to spawn for weapon, esp for for helmet...). And if you want the thing that's being squelched, you can turn the squelch for it off. In particular, any ego that you might see on a character very deep in the dungeon putting his kit together shouldn't ever be auto-squelched, because even if you don't want it now and you have several swap pieces that cover it you never know when you'll need to re-arrange EVERYTHING to fit in another flag (I've gone through this a bunch). But no one rejiggles their inventory to fit in infravision or slay orc before going to kill Sauron - at least not in vanilla angband.