Learn it → Apply it → Save as Skill
The fastest way to improve something is often to remove what is hurting it, not to add something new. Knowing that a habit, clause, feature, or ingredient is harmful is usually more reliable than knowing that a new one will help, so subtraction deserves the first look, not the last.
The Dr. Wallace’s Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
It is usually easier to know that something is wrong, harmful, or unnecessary than to know what new thing would actually help. A diet that removes a harmful ingredient rests on firmer ground than one that adds a trendy supplement. Improvement often means clearing away what is broken, not layering something new on top of it.
Important: via negativa is not "remove everything." It is removing what is confirmed harmful or useless first, before reaching for something new and unproven.
The model is most useful whenever the reflex is to add a fix on top of a problem, and you want to check whether removing the problem itself would work better.
The method walks from naming everything present to cutting the smallest, highest-confidence set of harmful or useless items.
Write down everything currently present: features, steps, habits, clauses, meetings, ingredients, rules.
For each item, ask whether it is actively causing harm, adding no value, or just there out of habit.
Before cutting, confirm why the item exists in the first place. A fence still standing may be doing a job you can’t see.
Cut the smallest set of highest-confidence items first, then reassess before adding anything new.
Open a use case to see what the model changes.
List every screen, field and step in the flow. Find the ones that do not serve the user's core task and remove them, instead of layering a tutorial or tooltip on top of a confusing flow.
Cut redundant sentences, hedging phrases and sections that repeat the same point, rather than adding more explanation to prop up a paragraph that is not working.
Before adding a new supplement, workout plan or productivity app, identify what is already actively hurting your sleep, energy or focus, and remove that first.
Instead of offering something new to sweeten the deal, identify and remove the specific clause or condition the other side actually objects to.
Rather than adding hedges or diversification on top of a shaky position, identify and remove the single dependency or assumption most likely to cause ruin.
Before adding a new meeting or approval step to fix a coordination problem, find and cut the existing step that no longer serves the purpose it was created for.
This is an illustrative example designed to teach the method.
Expand for where the phrase comes from and how its meaning shifted.
"Via negativa," Latin for "negative way," comes from apophatic theology, a tradition developed by thinkers such as Pseudo-Dionysius the Areopagite and later the Jewish philosopher Moses Maimonides. It describes the divine by stating what it is not, since human language was considered too limited to positively define it. The original sense is about description, not decision-making.
Statistician and essayist Nassim Nicholas Taleb adapted the term in his 2012 book "Antifragile." He argued that subtractive knowledge, knowing what is harmful, is generally more robust than additive knowledge, knowing what helps, because a single clear case can prove something harmful while proving something helpful usually requires much broader and shakier evidence.
It overlaps but is not identical. Minimalism is often an aesthetic or lifestyle preference. Via negativa is a decision rule about where reliable improvement tends to come from: it argues for removing confirmed harm before adding unproven benefit, not for owning fewer things as a goal in itself.
The model calls for removing what is confirmed harmful or useless, not stripping a system down until it stops working.
Some redundancy exists to prevent a specific failure. Check what job an item is doing before assuming it does nothing.
Personal irrelevance is not evidence of uselessness to everyone else relying on the same system.
A new feature, rule or habit built to compensate for a problem often just adds more surface area to maintain.
Cutting several things simultaneously makes it hard to tell which specific removal caused the improvement.
Some situations genuinely require building something new, not merely clearing away what exists.
a product, process or plan genuinely lacks something essential that no amount of removal will supply.
you are unsure why an item exists at all. Understand its purpose before deciding whether to cut it.
the goal is to gain from disorder and stress, not only to remove sources of harm.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply Via Negativa, what questions to ask, what evidence to seek, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
No. It is a sequencing rule, not a ban. Check what you can remove with high confidence first, because that knowledge tends to be more reliable, then consider addition once subtraction is exhausted.
Risk aversion avoids new bets in general. Via negativa specifically distinguishes the reliability of two kinds of knowledge: what is confirmed harmful versus what is merely hoped to help, and acts on the more reliable one first.
A system that survives by removing known sources of fragility depends on fewer untested assumptions than one that survives by stacking new fixes on top of old problems. Fewer untested assumptions generally means fewer ways for a plan to fail in an unexpected direction.