Answering uv4's critique of HOFi Xousk D1

Just repeating my comments, go there to read his:

This playlist is called "casual" and isn't my attempt at optimal play. I started recording DCSS because I caught a virus and was too sick to focus on work, so just fired it up. I'm not really using Textmind here and haven't compiled my notes. However I am TAKING notes and playing in a way that's fun for me.

It's not useful to play optimally during experimentation, because it takes too long to get informative results about the current question. A strategic improvement will be noticeable despite tactical sloppiness.  

In short, I don't recommend or expect anyone to watch this, I'm just practicing. I have no intention to produce definitive DCSS video until I've recorded the Castle of the Winds run. And I'm busy doing Spacemacs code to support Textmind right now, so that's on hold as well.

I don't care what your intentions were in recording when I'm taking notes, because they're for my benefit. I share them only to elicit disagreement on the meta. I've often acknowledged that your videos don't necessarily represent your skill level when playing unrecorded. For example, you play worse when commentating, which is understandable, as entertaining requires a lot of effort. If I wanted to evaluate your skill, I would watch tournament or streak ttyrecs.

I don't mind at all you critiquing my videos. I'll be interested to read your comment soon. I'm just clarifying the context. These videos would be pretty silly if they were purported to be definitive. Rather, I'm documenting my ongoing learning curve developing the essential combos, and enjoying them in downtime. It was often alleged that my notes were generated without actual game contact, which these videos refute. Plus, I need practice recording to find a suitable style.

@ 8:12 Unaware of how long it takes to put on a helmet leading to the 1st enemy of the game having a completely free attack.

Of course I know that equipping armor takes 5 turns. He was 4 tiles away, and random energy means it's uncertain whether he would get a free attack.

The question is whether 1 hit is worth the 2 extra AC from the helmet. The monster was an unarmed goblin. An turn 1 HOFi (axe) does 1.9 aed and receives .25 ish aed (fsim lacks hundredths decimal), for 1.65 net. With the helmet, received damage drops to .2 aed. So damage dropped by 1/5 at most. Goblin has 4 HP, so he dies in about 2 turns. Goblin max damage is 3. HOFi HP is 19.

So, is it worth giving up .25 aed in order to cut incoming damage from .25 to .2 for the 2 turns it takes to kill the goblin? Nope: .1 damage saved minus .25 taken = -.15

However, the danger isn't the goblin. It's what he attracted with his shout. If there's an entrance vault swarm, then I'll have to retreat to the killhole and endure. An extra 2 AC makes a huge difference in that scenario, and is well worth .25 aed. If the rush is coming, the best time to equip the helmet is right when I did it. Play was optimal.

Of course, my commentary was rather inane and didn't adequately express all of that. Which is why I needed practice.

No explored territory to retreat to if free attack did significant damage.

There were two pillars visible and two more implied.

@ 13:45 Consistently failing to retreat far enough to completely attenuate combat noise. Constantly risking additional enemies being alerted to make multi-staggered-engagement fights. Totally unnecessary and avoidable risk. This lazy (or uninformed - impossible to tell) play

I don't think completely attenuating combat noise is necessary or desirable, for these reasons:

  1. Surplus movement without net regeneration advantage depletes food resources
  2. Soft noise is one of the safest ways to pull

How far one should back off depends on the size of the current pull. That said I'm sure I wasn't playing optimally here.

@ 15:53 Decides to run from elec whip kobold by going downstairs to D2. Very dangerous move because no retreat back up is possible. Could possibly meet worse enemies on D2 and then have to fight the elec kobold while injured. Doesn't even occur to him to shaft the kobold to D2 so that D1 can be fully explored and mined for max XP and consumables.

Disagree again. D1 is over 3/4 cleared and the kobold is controlled. The only threat he poses is that he two-shots me before I can kill him in melee. Setting him loose on an uncleared floor allows him to combine with other threats for a kill. HOFi is OP for D2 with 8 AC and a large shield. He's been skilled properly and thus can handle typical fights, and still has his Might potion

@ 16:03 Not checking safety of pillar properly. This pillar is of crucial importance because Leo is effectively trapped on D2 due to elec whip kobold on the floor above. Then goes on to willingly fight a worm with no safe pillars and no possible retreat upwards. Insanity. If forced to run, could meet a D2 adder while significantly hurt.

Nah. HOFi's weakness is no stealth. Preferentially exploring hallways limits pull size and attack surface, allowing his melee prowess to prevail.

@ 19:44 Going up the unknown stair when a known stair is only a handful of steps further. If more D1 exploration is desired, it's safer to take the known stair and walk through tested paths. The unknown stair risks being adjacent to the elec kobold + new monsters in unknown territory. At worst, the known stair risks being adjacent to the elec kobold but at least it will be known territory with no new monsters. Yep indeed. The unknown stair turns out to be right near where the elec kobold was. Lucky to be unpunished this time.

Wrong. The kobold will be camping the stair he saw me descend, because I haven't spent any time on the floor for him to wander elsewhere. To prove this, I entered wizmode, descended a D1 stair in view of a kobold, waited 1.6k turns on D2, went up, and he was standing right next to it.

This critique is disappointing so I'm going to start skipping ahead.

@ 24:45 Leo thinks he would need to use the flame wand to kill the elec kobold safely. Decides not to, presumably to conserve the consumable wand charges. Doesn't realise the kill is risk-free from range with a hunting sling.

Slings miss, and HOFi Dex is low. Slings are also lower damage per turn. Normally HOFi inaccuracy doesn't matter thanks to his high strength making hits devastating, but an elec kobold could potentially get adjacent due to a corner, slip in and 2-shot before the axe connects. XL1 HOFi untrained sling damage is 1.1 aed, accuracy 44%.

Could have a free elec weapon + whatever other items and consumables are on D1 + whatever extra XP is there. Early elec whip would be fantastic because it cross trains with axes which are already trained.

Yes, early elec whip is fantastic, but this HOFi is already totally OP for D2 thanks to large shield and helmet. The kobold is one of the few threats that could conceivably kill him, thus it's worthwhile to delay him a bit, if there's no pressing need for an elec whip.

@ 30:21 Moving towards the last-known location of the elec kobold after shouting and alerting it. Kobold is allowed it to get a free move, closing the gap and allowing less time to shoot from range. Mistake not acknowledged, possibly unnoticed.

Just being cautious. Now that the kobold has received an alert, he will start to wander, and that can be randomized. I didn't want to assume he'd come from the expected direction, so I recentered. Plenty of room to gun him down.

Flaming wand charge wasted. Sling would have sufficed.

This is a general principle of mine. In a potentially lethal fight, lead with my best shot, particularly if its strength is uncertain. This gives time to adjust strategy if the best shot turns out to be weak. Usually what happens is the best shot does a lot of damage, greatly reducing the variance of the fight. Which then allows the use of cheap means to clean up.

Because DCSS is about variance management, early resource spending is much more effective than late. Since essential characters rapidly accumulate resources, there's no reason not to spend them promptly when they can actually improve the odds of winning the game.

There were definitely things to criticize in this video. For example, I should've gone directly to the kill room I eventually used, to minimize the kobold's chance to forget about me and start wandering.

Anyway, the Xousk video is garbage. Exercise your vindictiveness on Galost Fiawk's Elf:1 video:

It's not significantly different than what the definitive version will be, except that he's overleveled for Elf:1.

That said, I can guarantee there will be mistakes in the rest of his run, since I'm inexperienced with the midgame and have barely played the late game.

Show Comments