Start with skills, not a single score
A course score can hide very different needs. Two students may both average 70 percent. One misses equations. The other misses geometry. Useful practice needs a smaller unit than the whole course.
Lodera connects each question to one main skill and sometimes supporting skills. Your answer updates the evidence for those skills. A supporting skill can receive evidence without becoming the reason that question was served.
The skill list comes from the course topic list. Students can select all topics or choose a smaller set. The queue only serves skills that have active questions in the bank.
The six mastery labels
Lodera shows six labels: Learning, Building, Getting there, Solid, Strong, and Sharp. They turn your recent work into a readable progress level. The same labels are used for individual skills and larger topic groups.
A label is a summary of practice evidence. It is not a predicted SAT or AP score. It also is not a permanent grade. New answers can move the estimate as the system gets more evidence.
Topic progress combines the skills inside that topic. An unstarted skill still counts as unstarted, so finishing one easy part does not make the whole topic look complete.
How a skill reaches the front of the queue
The practice engine gives each eligible skill a priority. Recent correct and wrong answers affect that priority. Lower mastery raises the need for practice. Time since a skill appeared also matters, so ignored skills can return.
A miss keeps a skill in view. Clean correct answers reduce its urgency. The queue protects skills that still need evidence and releases a target after enough clean results. A wrong answer on a supporting skill can also raise its future need.
The system does not always choose the single highest number. It uses weighted sampling among strong candidates. That adds variety while still favoring higher-priority skills. Recent question branches are also moved down so the same shape does not repeat at once.
How the next question is chosen
After choosing a small set of skill targets, Lodera asks the server for active questions tied to those skills. The request includes recent question IDs so the bank can avoid immediate repeats.
The engine also sends a target difficulty for each skill. Stored questions have difficulty values. The server favors a question near the target, while still respecting the priority order and adding some variety.
The next question is already loaded when possible. That speed is a product detail, not part of the learning model. The important part is that the question comes from the selected topic scope and a ranked skill need.
What changes after an answer
An answer updates that skill and your recent result history. The engine records whether the attempt was correct and how strongly the question covered each skill. It then ranks the queue again.
One answer does not prove mastery. The queue uses windows of recent results, not just the last click. It also keeps a protected set so a weak skill is not dropped after one lucky correct answer.
Students can override the broad scope by picking exact topics. That choice is saved through an API route. The adaptive system then works inside the selected set instead of silently pulling from the entire course.
What adaptation cannot do
Lodera cannot see your written work or know why you chose an answer. A wrong choice may suggest a common mistake, but it is not a diagnosis of your thoughts. The system cannot replace a teacher who asks follow-up questions.
Mastery is an estimate based on Lodera practice. It does not include your homework, class discussion, sleep, test stress, or every skill on the real exam. It cannot promise a score.
The queue also depends on the bank. It can only serve active questions tied to covered skills. If a template is weak or a skill has little content, adaptation cannot create trustworthy material out of nothing.
A useful promise, kept small
Adaptive practice should mean less guessing about what to do next. It should not mean that software knows a student perfectly. Lodera uses answer history, progress by skill, time since review, chosen topics, and stored difficulty to make a next choice.
That choice is a ranked guess that improves with evidence. The student still decides when to stop, when to ask for help, and which topics matter most.
References
This article describes Lodera's current question-bank and practice-engine code.