Information triage

The emergency room does not take patients in the order they arrive

A car accident victim with internal bleeding walks through the door at 9:14pm. A teenager with a broken wrist arrived at 8:47pm. A man with chest pains has been waiting since 8:30pm. An elderly woman with a mild fever checked in at 8:15pm.

No competent emergency department processes these patients in arrival order. The woman with the fever, despite arriving first, waits. The man with chest pains gets assessed quickly. The car accident victim goes straight to a trauma bay. The broken wrist gets stabilized and queued.

This is triage. The word comes from the French trier — to sort, to sift, to separate. It entered medical vocabulary during the Napoleonic Wars, when Baron Dominique Jean Larrey, Napoleon's chief surgeon, faced a problem that no amount of surgical skill could solve: there were more wounded soldiers than he could treat, and treating them in the order they fell meant that soldiers with survivable injuries died while he spent time on soldiers who were either fine without treatment or doomed regardless. Larrey's innovation was to sort before treating. Assess every patient. Categorize by urgency. Treat the ones where intervention would make the greatest difference first.

This was not a medical insight. It was a resource allocation insight. And the resource being allocated was not medicine — it was attention.

You face the same problem every day. Not with patients, but with information. Your inbox, your feeds, your queues, your notifications — they all present items in arrival order, as if every piece of information deserves equal access to your attention. They do not. Some information loses its value in hours. Some retains value for decades. Some requires immediate action. Some requires no action at all. Some will change how you think. Most will not change anything.

In the previous lessons, you built the components of an information processing pipeline. You learned that information is raw material for decisions (Information is raw material for decisions). You mapped the pipeline (The information pipeline). You curated your inputs (Input curation). You learned to process each item with a decision — act, store, or discard (Processing means deciding what to do with each item). You built a reference filing system (The reference filing system) and an action filing system (The action filing system).

All of that infrastructure assumes you are processing items. This lesson addresses a question the previous lessons did not: in what order should you process them?

Why arrival order is the wrong default

Every digital system you use presents information chronologically. Email: newest first or oldest first. Slack: message order. RSS: publication order. Read-it-later: save order. Notifications: arrival order. The implicit assumption is that the order in which information reaches you is the order in which it should receive your attention.

This assumption is wrong, and the consequences of following it are severe.

Arrival order is random with respect to value. The most important email of the week might arrive at 2am on Saturday, buried below seventeen promotional messages that arrive on Sunday morning. The article that would change your approach to a critical project might be the fourteenth item in your read-it-later queue, behind thirteen articles you saved impulsively and will never finish. The Slack message that signals a project is about to fail might be lost in a channel where 35 of the previous 40 messages were casual conversation.

When you process in arrival order, you give your freshest attention to whatever happened to arrive first, regardless of whether it deserves that attention. By the time you reach the high-value items — if you reach them at all — your processing capacity has been partially spent on low-value items. This is the informational equivalent of the emergency room treating the mild fever before the internal bleeding because the fever patient arrived first.

The fix is the same fix Larrey discovered on the battlefield: sort before you process. Assess before you treat. Triage.

The two dimensions of information triage

Medical triage sorts patients on two axes: severity and treatability. Information triage sorts on analogous axes: value and time-sensitivity.

Value is the degree to which a piece of information will improve a decision, enable an action, or change your understanding of something that matters. A client's revised project requirements have high value — they directly affect what you build next. A newsletter's summary of industry trends has moderate value — it might inform future decisions. A social media notification has low value — it is interesting for thirty seconds and changes nothing.

Time-sensitivity is the rate at which the information's value decays. Some information is like fresh fish: valuable today, worthless tomorrow. A meeting request for this afternoon. A breaking development in a negotiation you are running. A time-limited opportunity. Other information is like canned goods: equally valuable whether you process it today or next month. A research paper on leadership psychology. A book recommendation. A reference architecture for a system you will build next quarter.

Crossing these two dimensions gives you the Eisenhower Matrix, originally attributed to Dwight D. Eisenhower's observation that "what is important is seldom urgent, and what is urgent is seldom important." The matrix, later popularized by Stephen Covey in The 7 Habits of Highly Effective People (1989), divides all items into four quadrants:

Quadrant 1: High value, time-sensitive. Process immediately. This is the client email with a deadline, the system alert that requires a response, the time-limited decision that cannot wait. These items go first, always.

Quadrant 2: High value, not time-sensitive. Schedule for focused processing. This is the strategic article, the research paper, the long-form analysis, the professional development material. These items are where your greatest returns hide — they improve the quality of your thinking over time — but they never scream for attention because they have no deadline. If you only process Quadrant 1 and let Quadrant 2 accumulate, you will be perpetually responsive and never strategic.

Quadrant 3: Low value, time-sensitive. Batch process quickly or delegate. This is the routine reply, the administrative form, the FYI notification that expires if not handled today. These items feel urgent because they have deadlines, but they contribute little to your goals. Process them in a batch during low-energy time — do not give them your sharpest hours.

Quadrant 4: Low value, not time-sensitive. Discard or deprioritize aggressively. This is the newsletter you subscribed to out of guilt, the social media update, the article someone shared that is mildly interesting but irrelevant to anything you are working on. Most of what arrives in your information stream falls here. The triage decision is to recognize this category quickly and spend almost no time on it.

The discipline of triage is not in the framework — anyone can draw a two-by-two matrix. The discipline is in the willingness to let Quadrant 4 items go without reading them, to batch Quadrant 3 items instead of responding immediately, and to protect time for Quadrant 2 items even though they never feel urgent. Most people spend the majority of their processing time in Quadrants 3 and 4, because those items arrive frequently and are easy to handle. The result is a life spent efficiently processing information that does not matter.

Information half-life: a triage accelerator

Not all information decays at the same rate. Understanding the decay rate of different information types makes triage decisions faster and more accurate.

The concept of information half-life — the time it takes for a piece of information to lose half its value — provides a useful mental model. Different categories of information have radically different half-lives:

Minutes to hours. Traffic conditions. Weather forecasts. Stock prices. Breaking news. Sports scores. Social media trends. This information is valuable only at the moment it is relevant and becomes worthless almost immediately. If you do not process it now, there is no point processing it later.

Days to weeks. Project status updates. Meeting agendas. Current events. Weekly industry news. Tactical recommendations. This information has a usable window, but the window closes. A project update from two weeks ago is stale. A meeting agenda from last Tuesday is irrelevant.

Months to years. Technical documentation. Product specifications. Research papers. Industry analyses. Market data. This information remains useful for extended periods but eventually becomes outdated as the underlying reality shifts.

Decades to centuries. Principles of logic. Mathematical theorems. Foundational scientific findings. Philosophical arguments. Human psychology. History. This information has effectively infinite shelf life. A principle that has been useful for fifty years is likely to remain useful for another fifty.

This last category leads to a powerful triage heuristic drawn from Nassim Nicholas Taleb's concept of the Lindy effect. The Lindy effect, which Taleb explored in Antifragile (2012), states that for non-perishable things — ideas, books, technologies — the longer something has survived, the longer its remaining life expectancy. A book that has been in print for a hundred years is likely to remain in print for another hundred years. A book published last month has a much shorter expected lifespan.

Applied to information triage, the Lindy effect produces a counterintuitive rule: older information that is still considered relevant deserves higher triage priority than newer information that is merely novel. A classic paper on decision-making that has been cited for forty years is more likely to improve your thinking than a blog post published yesterday. The news cycle optimizes for novelty. Your triage system should optimize for durability.

This does not mean you ignore new information. It means you assess new information against a higher bar. When a piece of information arrives and you are deciding its triage priority, ask: will this still matter in a week? A month? A year? The newspaper test, as some call it — would this be on the front page of a newspaper next month? — is a rapid heuristic for estimating half-life. Most of what arrives daily fails the one-week test. Very little passes the one-year test. The items that pass the one-year test are almost always worth your best attention.

Warren Buffett, who famously spends roughly 80% of his working day reading, applies an implicit version of this filter. He does not read 500 pages of news per day. He reads annual reports, industry analyses, business histories, and foundational texts — material with half-lives measured in years or decades. His information diet is heavily weighted toward Quadrant 2: high value, not time-sensitive, long half-life. This is not accidental. It is triage.

The triage pass: operational protocol

Here is the concrete protocol for triaging your information before processing it. This takes between two and five minutes and should be performed at the start of every processing session.

Step 1: Survey the full landscape. Before opening a single item, scan the surface-level indicators of everything waiting for you. In email, this means subject lines and senders. In Slack, channel names and message previews. In a read-it-later queue, titles and sources. In a notes app, item titles and timestamps. You are not reading. You are scanning. The goal is to see the full scope of what is waiting so you can prioritize across the entire set, not just within the first few items you happen to encounter.

Step 2: Flag Quadrant 1 items. As you scan, identify anything that is both high-value and time-sensitive. These items get a mental or physical flag. Do not process them yet — just mark them. In a typical scan of 50-100 items, you will find between zero and five Quadrant 1 items. If you find more than five, either your inputs need stricter curation (revisit Input curation) or your threshold for "urgent" is too low.

Step 3: Identify Quadrant 2 items. Look for high-value items that are not time-sensitive. Strategic articles. Deep analyses. Important but not urgent communications. These will be processed after Quadrant 1, during a dedicated focus block. Set them aside.

Step 4: Batch-mark Quadrant 3 and 4. Everything remaining is either low-value-but-time-sensitive (Quadrant 3, to be batch-processed quickly) or low-value-and-not-time-sensitive (Quadrant 4, to be discarded or deferred indefinitely). Most items fall here. The key discipline: do not feel guilty about this. The purpose of triage is to reveal that most of what arrives does not deserve your best attention.

Step 5: Process in quadrant order. Quadrant 1 first, then Quadrant 2, then Quadrant 3, then Quadrant 4. Within each quadrant, you can process in any order — arrival order is fine as a tiebreaker when priority is equal.

The entire triage pass should take less time than processing a single Quadrant 3 email. If it takes longer, you are reading instead of scanning. The moment you catch yourself actually engaging with the content of an item during triage, stop. Flag it and move on. Triage is sorting, not processing. The two operations are fundamentally different, and conflating them is the primary reason triage attempts fail.

Speed versus depth: triage determines your reading mode

Triage does not only determine the order of processing. It determines the mode of processing.

Not every piece of information deserves the same depth of engagement. Some items deserve deep reading — slow, careful, annotated, synthesized. Others deserve scanning — fast, extractive, looking for the one or two relevant facts. Others deserve skimming — a few seconds to confirm there is nothing important, then discard.

Your triage categorization should set the reading mode:

Quadrant 1 items: focused processing. Read carefully. Extract action items. Respond thoughtfully. These deserve your full attention because both the value and the time-sensitivity are high.

Quadrant 2 items: deep reading. These are the items worth scheduling dedicated time for. Do not skim a high-value article during a quick processing pass. Flag it, queue it, and give it the reading time it deserves. This is exactly what read-it-later systems (the subject of the next lesson) are designed to support.

Quadrant 3 items: rapid processing. Batch these. Reply quickly. File quickly. Make the minimum viable decision and move on. Do not spend ten minutes crafting a response to a routine request. Two sentences are fine.

Quadrant 4 items: scan and discard. Three seconds per item. Subject line. Sender. Confirm it is low-value. Delete or archive. Do not open, do not read, do not engage. The cognitive cost of opening a low-value item and reading it is far greater than the cost of discarding it unopened — and the risk of missing something important is negligible, because you already surveyed the full landscape in Step 1.

This modal matching — adjusting your processing depth to the item's triage category — is where the real time savings emerge. Most people apply a single processing mode to all items: the careful, thorough read they would apply to a Quadrant 1 message. Applying that mode to Quadrant 4 items means spending two minutes reading a newsletter you are going to delete anyway. Across 50 Quadrant 4 items per day, that is nearly two hours of wasted reading time — more than enough to fund the deep Quadrant 2 reading that actually improves your thinking.

Emergency information protocols: when triage becomes critical

In crisis communication — disaster response, cybersecurity incidents, medical emergencies — triage is not optional. It is the first and most critical operation, because the cost of processing information in the wrong order is measured in lives.

The Federal Emergency Management Agency (FEMA) uses a structured triage framework for incoming information during disasters: categorize every report by geographic scope, affected population, time-sensitivity, and resource requirements. Information that indicates imminent danger to life gets processed before information about property damage, which gets processed before logistical information, which gets processed before administrative information. The triage categories are set in advance. When a crisis hits, responders do not debate priorities — they apply the pre-established framework and process accordingly.

The lesson for personal information management is not that your daily email constitutes a crisis. It is that the best time to design your triage criteria is before you need them. If you decide your triage categories while staring at a full inbox on a Monday morning, the urgent items bias your judgment — everything feels urgent when you are already behind. If you establish your triage criteria during a calm moment — deciding in advance which senders get Quadrant 1 status, which types of information are always Quadrant 4, which keywords signal time-sensitivity — the Monday morning triage pass becomes mechanical rather than deliberative. You apply the framework instead of inventing it under pressure.

This is a pre-commitment, and it connects directly to the commitment architecture principles from Phase 34. You are committing to a triage protocol in advance so that you do not have to make triage decisions in the moment when your judgment is compromised by the very overwhelm that triage is meant to solve.

The Third Brain: AI as triage assistant

AI is exceptionally well-suited for information triage because triage is primarily a classification task, and classification is what modern language models do best.

An AI assistant can perform the initial triage pass for you. Feed it your inbox — subject lines, senders, first sentences — and ask it to categorize each item into your four quadrants based on criteria you define. The AI does not know your priorities as well as you do, but it can learn them quickly from a few examples, and its classifications will be correct for the obvious cases — which constitute the majority. Quadrant 4 items are easy to identify algorithmically. Quadrant 1 items are harder, but an AI that knows your active projects, your key contacts, and your deadlines can flag likely candidates with high accuracy.

More specifically, AI can estimate information half-life. Given a piece of content, an AI can assess whether the information is time-bound (news, event-specific, market conditions) or durable (principles, frameworks, research findings). This assessment can automatically inform triage priority: durable, high-value content gets flagged for Quadrant 2 deep reading, while time-bound content gets flagged for immediate processing or discard depending on its value.

AI can also perform the speed-versus-depth matching. For Quadrant 3 items, an AI can extract the essential information and draft a response, reducing your processing time to a quick review and send. For Quadrant 4 items, an AI can provide a one-line summary so you can confirm the discard decision without opening the item. For Quadrant 2 items, an AI can provide a preliminary summary that helps you decide when and how to schedule your deep reading.

The sovereignty constraint remains: you define the triage criteria. You make the final call on Quadrant 1 items. You decide what "high value" means in the context of your goals and projects. The AI handles the mechanical sorting — the scanning, classifying, and pre-processing that takes human attention but does not require human judgment. This division of labor lets you spend your limited cognitive bandwidth on the items where human judgment matters most, rather than spending it on recognizing that the forty-seventh email is another promotional newsletter.

The bridge to queuing

You now have a meta-layer above your processing pipeline. Before you process, you triage. Before you decide act/store/discard on individual items, you decide the order in which items receive your attention. The triage pass takes minutes. The value it creates — ensuring that your best attention goes to the highest-value, most time-sensitive items, and that low-value items are processed quickly or discarded — compounds across every processing session for the rest of your life.

But triage reveals a specific tension. During your triage pass, you will regularly encounter items that are genuinely valuable — Quadrant 2 material that would improve your thinking, deepen your expertise, or inform a future decision — but that cannot be processed right now because they require extended, focused reading. A thirty-page report. A long-form article. A dense research paper. You cannot read them during a quick processing session, and you should not — because the triage pass correctly identified other items as more immediately important.

These valuable-but-not-now items need somewhere to go. Not your inbox, where they will be re-triaged tomorrow with the same result. Not your reference system, where they will be filed as if already processed. They need a queue — a holding area specifically designed for content that you have committed to engaging with deeply, on a schedule that respects both the content's value and your current priorities.

That queue is the read-it-later system, and it is the subject of the next lesson.

---

Sources:

1. Larrey, D. J. (1812). Memoirs of Military Surgery and Campaigns of the French Armies. Translated by R. W. Hall, Joseph Cushing.
2. Covey, S. R. (1989). The 7 Habits of Highly Effective People. Simon & Schuster.
3. Allen, D. (2001). Getting Things Done: The Art of Stress-Free Productivity. Viking.
4. Taleb, N. N. (2012). Antifragile: Things That Gain from Disorder. Random House.
5. Mann, M. (2006). "Inbox Zero." Google Tech Talk, July 2006.
6. Isenberg, D. J. (1984). "How Senior Managers Think." Harvard Business Review, 62(6), 81-90.
7. Hallowell, E. M. (2005). "Overloaded Circuits: Why Smart People Underperform." Harvard Business Review, 83(1), 54-62.
8. Federal Emergency Management Agency. (2018). National Incident Management System. Third Edition. U.S. Department of Homeland Security.