To solve those problems, I made a web service to help to make and review extracts. By registering the bibliographic information of the book when starting to read the book, we can make a leverage note with a link to the bibliographic information with one touch.

However, I failed in the design of the review mechanism, and it brings the pain in a review.

I was too inclined to the concept read the note repeatedly and concentrate it more. In the service, I defined the best action is to remove unnecessary information from the note and make it shorter. The second best is to add missing information. I defined to keep the same length is the worst action. By this design, the notes are more concentrated and improved in quality.

However, because I defined "to review" as "make it shorter or longer," the review process becomes painful. I made a "to review" list in the service. If a note has nothing to remove or add, it remains in the list. The pile of tasks harms motivation.

I should make the cost of review as low as possible. Editing is too expensive. Review in Anki needs only one click. As a result, we can review even in short gap time.

Also, sometimes the note in "to review" list seemed to have no value. I forget the context and why I think it is important. I should add a button to mark it reviewed with one click in the service, however I can only edit it or delete it. Even if I feel it worthless, to delete is accompanied by fear. Fear harms motivation. I postponed my decision, and the list became full of trash.

In the next section, I introduce the Incremental Reading. It seems a good improvement against my failure.