Date: Tue, 18 Nov 1997 20:07:27 -0500 (EST) Message-Id: <199711190107.UAA28092@locke.ccil.org> Reply-To: Rick Nylander Sender: Lojban list From: Rick Nylander Subject: Re: Using LogFlash X-To: Lojban list To: John Cowan X-Mozilla-Status: 0011 Content-Length: 1183 X-From-Space-Date: Tue Nov 18 20:07:29 1997 X-From-Space-Address: LOJBAN@CUVMB.CC.COLUMBIA.EDU So to make sure I understand you... Lojbab: >So to answer your question, look at your typical New Word percentages and >back figure from around 15 errors to get an ideal New Word pile size. So, upon examining the log file, I see that I average 12 errors from a new word lesson (note that data is from New Word Review mode). This would indicate setting new word lessons to 25? >Continue this until your Dropback pile generates an error lesson between 20 >and 25, and then set your maximum lesson size to that number. (That oughta take about 3 sessions :-) I take it you mean the "Maximum lesson size for other lessons is..." variable. My observation is that words seem to be chosen randomly from the dropback pile. This would decrease your chances of getting dropback words right, because it will become longer and longer between repeating the same word as the dropback pile grows (I tend to get a lot wrong there too). As the dropback pile increases (between the new word errors and the errors in every other pile), it would seem like it wouldn't take long before the dropback pile would be HUGE. Or am I mistaken in some way, and this is the way it's supposed to be? Rik.