Deep Processing for Research

Part of Deep Processing

Deep processing for research is one habit applied while reading: each source is placed in a picture of the field at the time it is read, and not later at write-up. A source is placed when the reader can say three things about it: what it supports, what it leaves out, and which question in the field is still unanswered. A source that has not been placed has only been stored. Careful annotation does not change that.

The picture that builds up over the sources also tells the reader when to stop collecting. Without the picture, the reader stops when tired, and not when the picture is complete.

Core takeaways

• Each source gets placed against the others while it is read. Placed means the reader can say what it supports, what it leaves out, and which question in the field is still empty.
• Overload at first contact with a field is normal. The repair is an organizing pass of ten to twenty minutes with nothing new coming in. More reading does not repair it.
• The cost per paper falls as the picture forms. The first orienting paper can take about four hours. By about the fifteenth, a paper takes about ten minutes.
• Weak writing usually reflects weak organization of the material. The fix is to reorganize what has already been read before adding sources.
• A broad picture comes before a narrow question. A thesis question set before the field is intelligible forces every source through a small frame.
• A general model is useful at the edge of a picture the reader already holds. Handing it the placement stops the skill from developing.

Why research forces the change

Understand-and-recall is the study habit of reading a source, understanding it, and remembering it. It carries a student through school and through most of an undergraduate degree. It can be stretched through a master’s degree by doing higher-order work only when a question forces it. The habit does not cause failure. Its cost is inefficiency, thin and disconnected insight, and no free time. The habit survives into research because it worked for fifteen years. In research the option ends, because the volume of sources is too high and the output has to be new.

Deep Processing is the parent concept: working information for meaning by comparing, judging, and connecting. Deep processing for research is that work applied to a whole field instead of a chapter. Bear Hunter System is the encoding loop that runs the same moves on a single source; here the same moves are pointed at a field. Syntopical Reading runs the same placement at book scale.

The six moves on each source

Each source gets six moves. Understand what it claims. Locate it against the other sources as supporting, challenging, or ignoring them. Compare its views and methods with theirs. Detect what it does not see: its assumptions, its measures, its framing. Evaluate its strength, its generalizability, and its relevance. Build something from the gap the sources leave together.

On the page, the instrument for those moves is a short chain of questions asked after each claim. What does it mean? What is its relevance? How does it fit what is already held? Who agrees, and who disagrees? What is not here at all? The last question only works once the picture exists. A gap can only be noticed once there is a picture to notice it against.

Staying at understand-and-summarize is tempting because it feels like progress. More reading without active placement produces more confusion. Coverage without structure adds to the stock of unplaced sources. The hours that buy expertise are organizing hours. Reading hours and writing hours do not buy it.

Overload and the organizing pass

Feeling overloaded by sources usually marks first serious contact with a field, before expertise has organized it. It does not mean failure. It means the real work has started. The shape is familiar: many papers, many perspectives, contradictory claims, unclear gaps, overload, and a pull toward summarizing and filing. The repair is active organization. More reading does not repair it. The overload is stayed with. Passive re-reading is a way of escaping it, and it is not used.

Once the loop is running, organizing time exceeds consuming time. Reading comes in a short burst and is deliberately stopped. The pass that follows is about ten to twenty minutes of thinking with nothing new coming in. Overload does not fall steadily. New information raises it. An organizing pass lowers it. More information raises it again. A genuinely new perspective resets the cycle instead of adding to it. Thinking time swings the same way.

Provisional groups get built even when they are rough. Claims get compared directly: who agrees, who disagrees, and on what specifically. Filing them as “different perspectives” is not a comparison. In an emerging or thinly worked area the gaps are large and whole perspectives are absent. In a mature, heavily worked field they are not, and hunting for them in the same way wastes the pass.

The feel-test is explaining the field to a colleague in eight minutes without notes. The eight minutes comes from a short conference presentation, which compresses a year or more of work and exposes a disorganized schema. It stays a feel-test. It is not a finding.

How the cost per paper falls

The system this method comes from teaches a curve for working a new field from roughly twenty sources. The first large orienting paper can take about four hours, because every idea in it is new. By around the tenth paper, a paper yields one or two extra details and no new ideas. By around the fifteenth, a paper takes about ten minutes: it is read, recognized, placed, and often identified as the study an earlier paper was referring to. The curve is also a checkable expectation. Paper ten costing less than paper one, inside a single project, is the sign that the picture is forming.

How long the habit takes to build

The same system’s teaching estimate of the price: roughly four to six months of committed practice before the loop feels automatic. It takes longer where practice is interrupted for months at a stretch. Some practitioners take two years or more. The variance comes from practice volume. It does not come from talent. The early pass is supposed to feel slow. The payoff is deferred until the network of placed sources is dense enough that a new paper costs little to absorb. Quitting in week three because it still feels slow mistakes a deferred payoff for a failed method.

The picture is the deliverable. An hour is measured by whether the picture got clearer. Pages covered do not measure it.

Writing problems start earlier in the chain

Weak writing usually reflects weak internal organization of the material. That is the usual case and not the only one. Sometimes the difficulty is procedural: writing or speaking as a trained skill, which is developed separately and slowly with feedback. That case is not a schema problem, and this method does not cover it.

The limitation sits earlier in the chain than the point where the difficulty is felt. The chain, walked backwards from the output: elaborated prose; the bullet scaffold; the order and flow of ideas; organized ideas; higher-order organization; a non-linear representation of the material; the underlying processing skill. The protocol is to describe the difficulty, locate it on that chain, and clear only the earliest limitation. A later one is not relevant until the earlier one is cleared.

Read against that chain, the common symptoms each have a location. A literature review with no narrative, a discussion that will not write, and a presentation that will not compress all sit earlier than the writing. Ideas in disorder mean the higher-order schema is weak, so the map is reorganized before more writing. Examples missing or vague mean lower-order detail is thin, so the work goes back to the sources. A methodology that keeps shifting means the question was narrowed before the field was understood. Every new paper creating confusion means the current map cannot absorb it, so the map is rebuilt before more sources are added.

A researcher in difficulty, asked to explain the field the way it already makes sense to them, usually cannot. The system this method comes from puts that at around nine times in ten, as a teaching observation and not a finding. The reason is that the field does not yet make sense to them. The output cannot be high quality when the source is not, and the researcher is the source.

Repair usually comes from reorganizing what has already been read. Adding sources does not repair it. The early steps on the chain are the hard, high-value ones. The late steps are easy and low-value. Difficulty felt late comes from an early step done badly. It costs wasted time and misdirected attention.

From sources to a written piece

The path from sources to writing has a step that is easy to skip. Linear sources become a non-linear map built for learning. That map becomes an organized schema. Then a second non-linear pass, built for planning, unpacks the schema in the form the writing will take. Then a bullet scaffold. Then elaboration into prose, with local reordering as the writing goes. The second map is not the same artifact as the first.

Broad before narrow

Early narrowing feels efficient and slows the work. A specific question set before a broad schema forces every source through a small frame. A placement question is fine. A thesis question set before the map exists causes the slowdown. Once the field is intelligible, a narrow question becomes productive. Before that point it only shrinks how much of the confusion is in view.

The sequence: a few big-picture systematic reviews first; then deliberately contrary perspectives; then the recurring methods, authors, and gaps; then a provisional map; then the focused question; then the design or the argument. Prestudy is the same overview-first move at the scale of a session. This sequence runs it at the scale of a field.

Who to read in a field

A field may hold hundreds of thousands of researchers. The ones pushing its boundary are usually about five to ten, in the working model of a literature this method uses. A large remainder read that work and run small follow-ups or replications in different settings, which adds little at review scale. Reading the boundary authors’ publication histories covers most of what matters. Following those authors’ citations usually matters more than following citation counts.

Everything produced is a readout of the picture’s current state. That is why the repair is earlier in the chain, and why the question cannot narrow before the picture exists.

What a body of evidence can and cannot see

A hierarchy of evidence is useful for orientation. It does not replace thinking. Anecdote and case report sit at the bottom. Above them sit observational and cohort studies, used where an effect cannot be produced deliberately or where producing it would be unethical. Above those sits interventional work, with the double-blind randomized trial as the standard for primary research: participants do not know which arm they are in, and neither do the people measuring, so the expectations of either group cannot move the result. Above that sits non-primary work: meta-analysis and systematic review.

A single trial with thirty participants cannot support a claim that transfers to any meaningful population. Pooling about a hundred such trials can reach about four hundred thousand people, and that is why the pooled result is stronger. Pooling cannot add a measurement nobody took. A systematic review is the best available evidence up to its date, seen through the framing of the people who ran it. The thinner the field, the more that framing shows, because the reviewers may never have considered a perspective, and the data they examine is examined through the perspective they had.

The system this method comes from reads its own field this way, as a hedged reading and not a settled one: for roughly four decades up to about 2010, the literature on spaced repetition overwhelmingly tested short word lists at short delays. Very little of it measured the time the schedule costs, which learners can sustain it, or performance on higher-order testing across long retention intervals. The evidence base read as overwhelming because the question it answered was narrow. The last decade of work reads it down considerably.

The questions to ask of any review: What was not measured? Absence of evidence is not evidence of absence. Which assumptions made the research easier to run and less true to the setting it is applied to? Population, time cost, and transfer constraints are the usual ones. The working stance is to assume the gap is there and look for it, because a gap the reader does not believe in will not be seen. Every review has one.

Research thinking begins at the move from “what does the literature say” to “what can this literature see and not see”. Applied Critical Thinking: Testing Frames covers how to question what a source cannot see.

Using AI without handing over the placement

A general model is most useful at the edge of a picture the reader already holds. It can bring in material beyond that edge. It should not hold the map. If it does the placement better than the reader does, that says something about the reader’s current skill. Each time placement is handed to the model, the skill does not develop, so the overload never resolves and the dependence deepens. The reverse also holds: someone who already has the schema gets more from the same tool and can tell when an answer is wrong. The Right vs Wrong Way to Work With AI holds the general boundary rule. The research-specific uses follow.

A general model’s picture of a field is bounded by what it can retrieve. A large share of journal literature sits behind paywalls it cannot pass, and when it hits one it goes elsewhere. So its account of a field skews to the most mainstream and most accessible material, and where material is missing it produces confident, plausible citations that do not exist. The cause is structural. It is not a capability gap that the next release closes, which is why the rule outlives any model. Fabricated citations are a documented failure mode. They are not a permanent scoreboard.

A citation-network mapper suggests papers to read and generates no claims. A general language model generates claims. The mapper is compatible with the rule, because placement stays with the reader, and it is faster than walking citation lists by hand. No product is required.

Good uses sit at the edge of a map already held: check the map for gaps; get keywords for a gap already identified; surface contrary positions; find weaknesses in the current explanation; name prominent authors and concepts to go and read. The highest-value use runs in the reverse direction: state the current understanding and ask what perspectives it is missing. That returns terms to go and read, and it can save weeks of unfocused reading. In a genuinely unfamiliar domain, a short generated orientation can serve as a starting point only, before any review is read.

Poor uses hand over the placement: summarize this field, write the literature review, tell me the consensus, explain the best theory. Separately, reading a generated summary before the paper sets the frame the paper is then read inside. That holds even when the summary is accurate.

Authors are usually glad to send a copy of their own paper on a polite request. The journal pays them nothing for it, and they want it read.

The picture holds what the field says and what the field is able to see. It is the one artifact that cannot be held outside one head. A model trained on that literature inherits its blind spots, and then loses the paywalled part on top.

Keeping references without losing the map

Non-linear encoding is built for understanding and evaluating material. It does not preserve attribution the way linear notes do. That is a property of the technique and not a mistake in using it. The answer is a parallel system for references, run alongside the encoding.

Three strategies exist, chosen by use. Combinations are normal.

Synthesise as you go. After every two or three sources, write a short referenced synthesis, then keep adding to, modifying, and rearranging that piece as more sources arrive. Two parallel sets result: a non-linear map carrying understanding and judgment, and a linear cited draft that consolidates the learning and doubles as the reference bank. This works best when one piece of writing is the target. It is weak when the same references will be reused across projects and years; there it is combined with the second strategy.

A second brain. One page per source, with a short summary and deliberately chosen tags, so that the tag network shows relationships in a graph view. This is a tool class, local notes with tags, and no particular app is required. The setup cost is real. Trial the structure on a handful of references before committing, because a large base built on a bad template is very hard to change afterwards. Knowledge Base as Thinking Partner is the sibling page where the page-per-source system is worked out in full.

Both of those strategies extend study time and add administrative work. That is the cost of tracking a high volume of references that must be reused.

Reference chunking. Sources are filed by the job they do, never by topic or date. The categories: core foundational works, cited in almost any deep discussion; high-leverage applied studies with direct, frequent use; contextual works that broaden understanding but are cited rarely, sub-categorised by theme where that helps; controversial or counterpoint works that challenge the prevailing reading, a category that is not always viable where competing schools are the discipline’s foundation; and methods and measurement references, kept for templates, statistical approaches, and measurement tools. The mechanics: tags for filtering, a note on each reference saying why it sits in that category, and collections per project. The payoff has three parts: citations assemble fast, important but rarely used works stop disappearing, and the strong and thin places in the base become visible at a glance.

Filing by job requires enough grasp of the field’s principles to judge importance and relevance accurately. So chunking is the last of the three to become available. It is not a more advanced version of the other two.

Citing the right reference depends on a relational understanding of the topic and on frequent use of references in writing. That frequent use is a natural form of interleaved retrieval. Spaced Interleaved Retrieval is rebuilding the field from memory instead of recognizing papers, and frequent citation use is that same retrieval under another name.

What it feels like when it is working

The good feel is increasing command over a messy territory. It is not less work. The difficulty is better organized, and the confusion has changed character without disappearing. The swing described above continues: information raises the load, and an organizing pass lowers it.

Papers fall into camps. Names and dates become memorable without brute force, because the chronology is causally connected: a finding published in 2011 could not have appeared in 2015 once what came out in 2013 is known, and that connection is what makes it hard to forget. Gaps get specific. Writing finds a natural order. Questions sharpen because the field demands it, and not because one was chosen. The field starts to feel arguable. The checkable sign: new papers slot in and consolidate on contact, so reading happens in the gaps of a day and no longer needs a dedicated block.

Warning signs and limits

Warning signs: summaries that stand alone; notes growing while the map stays vague; generated summaries that feel clearer than the reader’s own understanding; a question narrowed before the field was intelligible; more reading producing more confusion; writing started before the schema can explain itself without notes.

Three limits. The schema-first rule can become an excuse never to start writing. The four-to-six-month ramp does not pay on a short, bounded project, where a straight collect-and-write is the correct call. And sometimes the difficulty is procedural writing skill, which none of this touches.

The filing systems serve the picture. They do not replace it. The field is organized when a new paper costs ten minutes and lands somewhere that can be named out loud with everything closed.

How to practice this

1. After each source, say what it supports, what it leaves out, and which question in the field is still empty. Notice whether you can answer all three. A source you cannot place has only been stored.
2. Read in a short burst, then stop. Spend ten to twenty minutes organizing with nothing new coming in. Notice the overload drop after the pass and rise again with the next batch.
3. Compare claims directly: who agrees, who disagrees, and on what specifically. Notice which papers fall into camps and which gaps get specific.
4. Track the time each paper costs inside one project. Notice whether paper ten costs less than paper one. That drop is the sign the picture is forming.
5. Explain the field to a colleague in eight minutes without notes. Notice where the explanation stops making sense. That is where the schema is still disorganized.
6. When writing stalls, describe the difficulty and locate it on the chain from prose back to processing skill. Clear only the earliest limitation. Notice whether reorganizing the map fixes the writing before any new source is added.

Open questions

Where does the line actually sit, in a live project, between a placement question and a thesis question?

What does a research field look like when it is mature enough that gap-hunting stops paying?

Sources

Adler, M. J., & Van Doren, C. (1972). How to Read a Book (rev. ed.), ch. 20. Syntopical reading: the point is the conversation, not the book.

Hart, C. (1998). Doing a Literature Review. Sage. The review as an argument, not a pile.

Booth, W. C., Colomb, G. G., & Williams, J. M. The Craft of Research. University of Chicago Press. Research as a conversation with a claim.

Nestojko, J. F., Bui, D. C., Kornell, N., & Bjork, E. L. (2014). Expecting to teach enhances learning and organization of knowledge in free recall of text passages. Memory & Cognition, 42, 1038–1048. Preparing to explain as a test of organization.

Altman, D. G., & Bland, J. M. (1995). Absence of evidence is not evidence of absence. BMJ, 311, 485. The measurement gap.

Ioannidis, J. P. A. (2005). Why most published research findings are false. PLOS Medicine, 2(8), e124. Reviews inherit a field’s framing, measures, and populations. GRADE and ROBIS are the later instruments for that inheritance.

Walters, W. H., & Wilder, E. I. (2023). Fabrication and errors in the bibliographic citations generated by ChatGPT. Scientific Reports, 13, 14045. Fabricated citations as a documented failure mode, not a permanent scoreboard.