It's not a bug, it's a feature! - FAQs

Frequently asked questions

Q: The same question keeps being repeated. Bore!
Q: At Lesson 44 force-scheduling stopped working:
Q: In "YOUR RESULTS" speeds are definitely too big

Q: The same question keeps being repeated. Bore!

A: In general whenever a new note is drawn it is chosen NOT to be the same as previous question. This keeps things more interesting. The only exception is when there are some forced-scheduled notes.

Forced-scheduling appears whenever a student makes a mistake. The note which he erred about is scheduled to be the next second, fifth and eighth question. If this note is drawn at random for e.g. the next 4th question it will still be repeated (due to force-scheduling) as the following (5th) question.

If one mistakes multiple times, and if 2nd, 5th or 8th question is already scheduled, new forced questions are scheduled for next available slot. Therefore, if one makes multiple mistakes about the same note it can be scheduled for few following subsequent question.

INBIF: what do you do if, learning a poem by heart, you keep saying "cage" instead of "gauge"? Well, you probably say "gauge, gauge, gauge, gauge..." a few times. This is exactly what forced-scheduling does for you if you keep making mistakes about the same note.

Q: At Lesson 44 force-scheduling stopped working:

I made a mistake but the note is not repeated!

A: Remark: The effect described here is always present, but it can be seen best in lessons with huge number of testnotes, and that's why I choose Ex.44 (with 23 testnotes!) as an example.

Each TEST starts from one series of force-scheduled questions to ask one question about each note (in random order). Only then random notes draw and new scheduling starts.

This means, that for lessons with huge number of testnotes you will not see any force scheduling or intelligent questioning effects until this first series ends. If, in Lesson 44, the first question asks about the 'G1' note, and you make a mistake, 'G1' will be scheduled to be ... 24th question (!), because the first 23 slots are already taken by the first run.

INBIF: changing this would require giving up the first, force-scheduled run of questions. I really don't want to do this. If there are two notes about which the student already answered, program will draw the slower-answered note with greater probability than the faster-answered note. On the other hand, if one note was answer and another not there is no way for the program to decide which of these notes should be more probable for future questions (*). That's why I decided that each test should start with one run of all testnotes in random order.

It really doesn't matter for lessons with e.g. four testnotes (if you make a mistake, the note will be simply scheduled as the fifth question). For lessons with big number of testnotes N, keep in mind that all the nice intelligent teaching features start only after the first N questions. This is one of reasons why default number of questions in a test is greater if a number of active testnotes is bigger.

(*) remark: well, in fact it could be done: program could record user's overall-speed from previous tests; and each new asked note could be assigned probability proportional to a fraction of it's speed to the overall-speed, while all not-asked-yet notes could be assigned some constant number instead. But this doesn't seem reasonable: should previous tests really influence note-draws for future? What if the student started form too easy lessons? What if he started from a too hard one? What if he practiced previous lesson a lot and gained good speed, and now moved to some new stuff? No, I really think, the way it is done now (each test starts from one force-scheduled run of all testnotes in random order, which can then be used as a gauge for probability for next draws) is more reasonable.

Q: In "YOUR RESULTS" speeds are definitely too big

A: No they aren't.

If you measure time of TEST, you measure both time you spent answering and "pauses" between questions. This leads to longer time and that's probably why you expected smaller speed. Well, this is not what speeddisplay shows. The speed shown there takes into account only time you spent answering (excluding pauses between questions).

Example: If you answer 80 questions, and on average between the moment when a note is shown and your answer there is one second, then your speed is

speed=1answ/sec = 600 answ/(10 min) (and this is what would be displayed)
 = 80 answ /(1' 20'')  

But the test will NOT take one minute and 20 seconds! It will last approximately two times longer(ca. 3'), because there is one-second pause between questions.