In an online experiment, screen-reader users answered questions about online charts correctly 34% of the time, against 87% for the comparison group; when the chart came with a data table, the screen-reader group rose to 73%. Only 1 of this site's 192 mermaid diagrams in Chinese posts has an accessible title. The only UDL meta-analysis I could read in full pooled g = 3.56, but most of the studies it included had no control group.
College-classroom meta-analyses find about half a standard deviation, but they compare whole-course teaching formats, the active sections usually include group discussion and instructor responses, and most controls were not done or not reported. In online courseware, "students who do more practice score higher on quizzes" recurs across a dozen-plus courses, but all of it is observational. None of the studies read here compared one person reading an article with and without practice.
In a quasi-experiment in one intro course, sections whose worked examples labeled what each step was for did better on quizzes right after learning, with no significant difference on later exams. Live coding, and Parsons problems in place of writing code, showed no measured difference in learning. Public GitHub notebooks rerun in their saved order finished without error about a quarter of the time; no study has measured how code that does not run affects what readers learn.
Large-sample records of "did anyone answer when the question appeared" come only from MOOCs: for a question shown in the player when the video reaches its position, about 74% of people who started that video answered; for pop-up review questions, about 70% of multiple-choice and about 40% of type-a-number questions were attempted. Under random assignment, about half as many people did a task that required typing as did one that only required ticking a box. None of the studies I read measured readers of public articles, and this site's traffic statistics are aggregate and cannot show who answers.
MOOC learners who posted in the forum passed at much higher rates than those who did not (57% vs 8% in one course), but the authors themselves call this a correlation under self-selection. Randomly adding a reputation system made replies more numerous and faster, with no difference detected in grades; an instructor replying actively versus not at all showed no difference in completion or engagement. No study has tested what a blog comment section does for readers' learning.
No study compares the same content as an interactive article and as a static one. When interaction was added to the same chart, none of the variants in one set of experiments beat the static version. What helped was readers doing something with the figure, and the gap disappeared once the comparison group also did something. Readers use the component that responds at a touch and sits in the middle of the narrative; about half of desktop readers never opened content placed behind a click. One long-time maker of interactive tutorials says each takes one to five months, including the time to learn the algorithm.
In university courses taught in English, subject-matter results showed no detected difference from teaching in the first language (d = 0.13). With same-language subtitles on foreign-language video, three experiments all failed to detect better comprehension; first-language subtitles came out ahead in one experiment and showed no difference in two. When scrolling was needed, comprehension on paper beat computers and tablets; there are only 3 direct comparisons of phones with paper. No study has measured how much non-native readers learn from web teaching articles read on a phone.
Peer interaction beats other learning conditions by an average g = 0.40, more when peers are told to reach consensus, but the participants were 4 to 18 and most studies had a single interaction; in Peer Instruction, 91% of students who picked the same wrong answer as their neighbor stayed wrong. Groups recalling together produce less than the pooled recall of the same number of people working alone; no study has compared adult self-learners reading alone with reading together.
Only expecting to teach averages g = 0.41 in samples aged 18 and over; no effect was detected in under-18 samples or among Japanese students. Expecting to teach and then writing an explanation or recording a video for an absent audience averages g = 0.48; teaching face to face is 0.84, from very few comparisons. A meta-analysis of creating teaching materials for others averages 0.17, with few studies of written materials. All of this comes from short lab materials; no study has tested writing a public article for strangers.
With no intervention, readers' ability to tell which text they understood and which they did not averages a correlation of only 0.18, which the authors convert to about 55% accuracy. Writing a summary or keywords after reading everything, or answering practice questions before judging, makes judgments somewhat more accurate; just reading ready-made questions and answers showed no detectable benefit. Most of the evidence comes from college students reading expository texts in the lab.
In one refutation-text meta-analysis, effects showed no measured difference whether or not studies first screened out readers who already answered correctly. A misinformation review lists 11 studies of the familiarity backfire effect; 3 reported that repeating a myth made people believe it more, all three on vaccines and all with single-item measures. No study has directly tested what happens when the reader never heard the misconception before. What you can act on now: keep the false claim and the correct claim together, give the correct claim with a reason, and do not let the false claim stand alone as a heading.
Jargon makes general readers rate a text as harder to read, and definitions that appear only on mouseover did not bring that rating back up. In studies of foreign-language reading tested right after reading, readers with glosses learned about 45% of the unknown words, versus about 27% without. Papers with more equations get cited less, and people have computed this correlation for both biology and physics, but the two sides never agreed on whether it holds up, and no study has measured whether readers understand less.
In a pooled analysis of 50 homework experiments on ASSISTments, hints, breaking a problem into smaller steps, and worked examples beat showing only the answer by d = 0.16–0.27, measured on performance during the homework itself. Interventions that worked in a few courses were scaled up to 247 MOOCs, and none of the main hypotheses replicated. Most course-quality rubrics don't publicly report reliability or validity; in one university's test, none of nine dimensions had a significant direct path to pass rates. The reviews of design-based research all look at how reports are written; none compares whether the materials it produces are more effective.
People who know more beforehand still know more afterward (correlation .53), but the average correlation with how much they learn this time is close to zero, and individual studies range from strongly negative to strongly positive. High-assistance materials help learners with low prior knowledge (d = 0.51) and hurt those with high prior knowledge (d = −0.43); high and low are relative splits within each study. Stating prerequisites at the top is industry practice; whether readers can use it to decide if a post is for them has not been tested.
In classroom studies, having students solve problems before instruction raises conceptual knowledge and transfer by about a third of a standard deviation on average, but every one of those studies had a teacher in the room; versions with group work had larger effects, except in interventions lasting only a few hours. A set of experiments that had roughly ten-year-olds solve first found the opposite. Unassisted discovery loses to explicit instruction, and discovery with scaffolding or feedback does better; no difference was detected between fading and not fading scaffolds. No study has tested one person reading an article and doing a project alone.
Compared with commercial textbooks, open textbooks showed no reliable difference in learning performance across 22 studies combined (g = 0.01), with fewer withdrawals but weaker evidence for that. Among four state-approved math textbooks in California, one scored about 0.05 standard deviations higher than the other three combined. Materials with a minimalist design (mostly rewritten manuals) had a large effect, but the studies are old, the samples small, and the meta-analysis did not go through journal review. The case that one-to-many explanation pays off and Diátaxis's four-way split are both arguments, not data; no study has directly tested re-explaining a course.
In a crossover experiment in a Harvard physics course, the class in which students scored 0.46 standard deviations higher on the test was the one they rated 0.56 lower; in an experiment with generative AI available, self-ratings rose more than actual improvement could explain. Viewing time, platform activity, and models predicting a pass mostly measure how much was done, not learning measured separately. This site's traffic statistics are aggregate, and none of its signals can measure whether readers learned.
In a randomized experiment in one AI course, an order built from rated transition difficulty beat a historical order by about 5 percentage points. Among 8 studies that sequenced interdependent content with algorithms, none significantly beat a fixed human-designed order. Practicing four problem types mixed across eight worksheets gave 61% versus 38% on a test about a month after review, a combined effect of interleaving and spacing. Spiral curricula have almost no efficacy research. MOOC learners who earned certificates still skipped 22% of the content on average.
Spreading practice across days shows an effect of about d = 0.54 in a meta-analysis of real classroom studies, which its authors consider smaller than a meta-analysis that mixed in lab studies. Spacing embedded in mathematics courses (which also involves interleaving) is g = 0.24, and when eight university STEM courses used the same procedure, the gain averaged only 1.5 percentage points, with no detectable difference. Rereading a week later showed no detectable advantage a week after that, and none of the studies in that classroom meta-analysis let learners decide for themselves whether to come back and review.
The most complete attempt, Quantum Country, embedded 112 questions in an essay that takes about four hours to read and then emailed readers to schedule reviews. Its authors rate the evidence that readers would remember much less without review as only emerging, and say there is no evidence it generalizes to other domains. When a MOOC pushed review questions to learners, about 80% of those assigned never touched them, and grades showed no detectable difference; the example that worked was a classroom experiment where review was scheduled into class time.
Telling readers what something is good for makes them rate the material as more useful, but across seven experiments no difference was detected in immediate test scores; in classroom studies where students wrote about usefulness themselves, results flipped across students and courses. Interesting details unrelated to the main point lowered immediate test scores on average, by an uncertain amount; only one pair of experiments detected no harm when learning was voluntary.
In HarvardX and MITx courses, about half of registrants on average never entered the course; among those who did enter, only 3.13% passed in 2017–18. A prompt to write a study plan raised completion from 13.8% to 17.7% in three courses, but at 247 courses no difference in completion was detected. These are all MOOC numbers; this site's traffic statistics are aggregate, with no data on individual readers' behavior.
Adding a graphic raises reading comprehension by about 0.39 standard deviations on average. Placing each diagram next to the text that explains it, and having readers do something with a diagram rather than just look at it, each have research behind them and cost less than drawing one more diagram, though no experiment compares them directly with adding a diagram; only one experiment isolated prediction, and it found no detectable difference. Use a state table for step-by-step execution, a numbered list for a single-path flow, and a table for comparisons. On a phone, a mermaid diagram with more than 3 nodes side by side in one rank shrinks until the text is unreadable.
Put a video at the top if it covers the whole article, at the start of a section if it covers only that section; videos at the bottom often get skipped. Readers mostly rewatch specific segments, so every section should link to its timestamp and the intro should state the video's length. Prequestions and writing your own explanation after each segment have experimental support, but those experiments all used short material, a little over 20 minutes at most.
Three practices have meta-analytic support: give readers a fully worked example, ask them to explain for themselves, and state a common misconception before refuting it. Questions before reading mainly help the content they ask about, and most experiments used math and science materials, short texts, and college students.
Asked content gains the most, according to research on classroom quizzes and prequestions (I found no synthesis of inserted questions themselves with a legally available full text), and feedback works better the more information it carries, though that evidence comes from classrooms, computer-based materials, and short laboratory texts. Readability formulas explain only a quarter to a third of teachers' difficulty judgments, and 30 Bloom verb lists disagree on where the same verb belongs.
Research interfaces share one move: keep the video and the matching text visible together, either beside the text or right at the passage being read. In the 16-person Papeos study, the condition with clips attached to passages had the lowest mean mental demand (the omnibus test across three conditions was significant). No experiment has compared layouts for a long written guide paired with a lecture recording, and no research I found says what to do on phones.
Pagefind scans static HTML after an Astro build and ships its index with a WebAssembly search runtime; the browser fetches only the index chunks required by a query, so no search server is needed.
Using my own 30+ RAG/Agent posts to audit the blog itself, I identified a prioritized improvement list spanning content quality, site tech, RAG design fixes, harness infrastructure, and AI agent applications — no phases, just priorities.
AEO (Answer Engine Optimization) is a content strategy aimed at AI search engines like Perplexity, ChatGPT Search, and Google AI Overview. The core idea is to make your content the easiest source for AI to cite — not just another link in the results page.
SEO is more than keywords. Structured data (JSON-LD), Open Graph, hreflang, and robots.txt are the technical optimizations that actually help search engines understand your content. This guide walks through a complete implementation using an Astro blog as the example.
To consolidate scattered notes and showcase diverse interests, I built a personal blog using Astro + Cloudflare Workers D1, paired with a Claude post skill for zero-friction writing.
Switching templates means replacing the entire project foundation. Figure out what you actually need first, then choose between AstroPaper, Cactus, or AstroWind.