One of the shibboleths of DCSS is that tactics are much more important than skilling. This is because the game is not hard enough for elites to agree on skilling, but a tactical failure that leads to death is obviously wrong. They're quite proud of this shibboleth, eagerly scoffing when it's contradicted. Ironically, this merely demonstrates that their game understanding is systematically superficial.
In fact, there's no objective way to compare the importance of skilling versus tactics. What degree of tactical incompetence is equivalent to skilling incompetence? One could arbitrarily say that the lowest level of tactics is pure o-Tab, and the lowest level of skilling is default settings. In that case, the literal idiot player, who is worse than a bot, will charge forth to his death on D1 every time, due to failure to rest.
Is this a meaningful or insightful comparison? Does it have any relevance to actual players? No. Actual players can apply common sense to cautiously explore the Dungeon, because humans are born with easily-awakened hunting instincts. However, humans have no applicable instincts about the skilling system.
Good skilling is the key that unlocks the smooth functioning of a build. It allows one to calibrate when a situation deserves consumables, when to halt an operation and attack elsewhere. It makes common-sense tactics effective.
Good skilling is the key to growing as a player. One's instincts about what should be possible, how difficult something should be, will be inaccurate until one learns the simple skilling best practices. One will absorb all kinds of false lessons and preferences that are really merely artifacts of poor skilling.
The DCSS community lacks even the most basic conceptual framework for discussing skilling. A sequence of skill level targets, sometimes tied to XLs, is as good as their advice gets.
The DCSS community has rejected my innovation to solve this problem, even though it is obviously superior. The relevant number when making a skilling decision is marginal cost. Some skills are worth more per level than others. One should pay more for high-value skills. Call the highest-value skills "primary". Call skills that are only half as valuable "secondary". Call skills 1/4 as valuable "tertiary".
Flexibly train skills, keeping the ratio of marginal costs roughly consistent with how valuable that skill currently is to the character. Develop an intuition, and discipline it with regular fsim experiments.
Note that in the early game, particularly the first few floors of Dungeon, skills are more granular. It's often better to view them as discrete levels rather than continuous marginal costs, particularly those skills with step functions: Fighting (mHP), Spellcasting (spell levels, mMP), Dodging, Armor, Shields, Invo/Evo (mMP). The bolded skills have no sub-integer impact.
The problem with the DCSS community's approach is that they continue to view them as discrete levels for the rest of the game. This is too rigid, fails to account for manuals and varying species aptitudes, and generally misses the huge impact of marginal cost. At higher XLs, this leads to ludicrous elite training decisions viable only due to DCSS' inverse difficulty curve and lack of difficulty setting.
there doesn't seem to be anything here