Quick Explanation of Method
Base scores are (presumed) always integers and cutoffs are found by multiplier*base (base = cutoff / multi). Play the song up to a score near the cutoff value, and note if the stars are present or not (its very important to do this step right, ive had multiple cases where i get the score and its different from what i expect, but i still note down what i expected). If possible you can get scores for multiple cutoffs on a single run. Keep track of the highest observed score that does not give the stars, and lowest observed score that does, and divide the bounds of each estimate by the mulitplier to get a decimal base score prediction. Then you take the highest 'low' base score prediction (across all cutoffs for which you have data), and the lowest 'high' base score prediction and those are your base score limits. If only one integer exists between those limits then you are done. I (Barfo) have an excel sheet which will graphically display this info, as well as give you working estimates based on current for what to aim at in order to narrow further. If you want to help out, PM me (or catch me on IRC, but i idle a lot) and we can start a conversation that will liekly end with you giving me your email, and me sending this excel file to it.

The width of the estimate obtained form proving a cutoff exactly is going to be proportional to 1/multiplier. So 1* and 2* cutoffs even if proven exactly, cannot give you the base score (you will get a range of 2-4 possible integer base scores). 3* sometimes can and mostly cannot, 4* always will and sometimes can prove the base even with a range of two for the 4* cutoff, and 5* always will prove it if you have it to within a range of 1 - 3 points and sometimes for a range of 4. 6* will have an even wider range but is generally not practical to shoot for

If you have good estimates, the best way to do it is to forgo 1 and 2* scores (too wide an estimate), and to palyt he song to within 500-3000 of the 3* score, then get the exact mod25 for either the lower or upper bound of the 3* cutoff, and then close out that score and note down the result. Most sogns will now have enough notes for you to playt o within a few k of the 4* score, and then get the mod 25 for the other side of the proof (if you got the lower bound of the 3* get the upper side of the 4*, and vice versa) and close out the score and note the result. Assumign you got both scores, In many cases you can plug those in and have the base score proofed. If you did not then you can replay and ignore the 3* cutoff, and go for the other half of the 4* proof, or some score pretty close to the 5* cutoff (if you make good use of SP and also prove your cutoffs fast is is possible sometimes to get a useful score for the 4* and the 5* cutoff in one run, but its tricky).

If you have a base score proven, enter it into the data table below, following the format , as well as whatever data that you gathered while pursuing the proof (per star cutoff: highest score not getting the stars, and lowest score getting them), even if some of that data is redundant (not the data used in either of the two limits for determining the base score). You do not need to fill on the entire table of course, but showing the data will allow it to be peer rewviewed, if needed. If you have only partial data that doesnt prove it then dont worry about making a line entry, except in pre-approved cases where some cutoff is so tough to do that we realize we need to work together on it (i find this case unlikely). The point of this table is as a record of work done to have a reference for Kawigi etc to easily see what data is present, to avoid duplication of efforts, and to allow for peer review of data, not to be a collaborative narrowing of individual songs (or a competition - the proof by line is to know who to go to for more info if errors are found). If down the line it seems there will be a problem with duplication coming about because of in progress proofs, then we can set up a system to reserve the proofs, but with 500+ songs, I do not anticipate this will be the case.

Top Priority Targets
Expert Guitar:

Song	Kawigi Estimate	raunch99 Estimate	spread
bodhisattva	44430	44404	-26
ourtruth	22189	22165	-24
lazyeye	59902	59889	-13
shootingstar	36423	36416	-7
americanwoman	30943	30939	-4
masterexploder	34048	34044	-4
goyourownway	34937	34936	-1

Current Data Table

Song	Inst	Diff	Kawigi est	Proven Base	1*	2*	3*	4*	5*	Proof By
Here It Goes Again	G	X	26619	26619	- / -	- / -	- / -	- / -	- / -	Barfo