# Talk:Blue Tongue

The description alludes to the dancing of a gravy fairy, which implies that this buff increases item drops. It would make sense, considering the effects of the other snowcones. ~IMJack 1/5 00:26

I spent 20 turns with blue tongue farming for pixels. I have no passive item skills and didn't have a gravy fairy type familiar equipped. During the 20 turns of blue tongue, I got 3 pixels **every** single adventure. As soon as the blue tongue ran out i went back to getting 1 or none each turn. Granted 20 turns isn't rock solid evidence, but it certainly does make me lean towards thinking it has something to do with +% item drops.--Williabr 16:55, 5 January 2006 (Central Standard Time)

I concur that the description implies a gravy fairy effect. --Darkmoon 02:50, 6 January 2006 (Central Standard Time)

I've always wondered how exactly you tell the exact item drop percentage of something. Banjooie 20:58, 9 January 2006 (Central Standard Time)

- What happens is the users test it for a certain amount of turns, say 1,000 or so, and compare the differences in various areas to tell the item drop percentage. --Wtf 21:27, 9 January 2006 (Central Standard Time)
- There are a few areas where the drop rates are pretty well established as well making this easier. --Veistran 17:03, 10 January 2006 (Central Standard Time)

I have done some research into the blue snowcones. My conclusions aren't yet specific enough to make the main page, but here are my preliminary results. In 940 adventures, using no other item drop boosters than blue tongue, I got 371 yeti furs. **This means that the item drop bonus is 32% with an uncertainty of about plus-or-minus 10%.** I think the most likely value for the true value of the boost is 25%, but it may be 30% or 40%.

This at least confirms that blue tongue *definitely* increases item drop rates.

Here's the maths - I'd appreciate if somebody reviewed it to see if I've got anything wrong.

The average drop rate was 371/940 = 39.5%. This corresponds to a boost of 31.5%, assuming that the drop rate for yeti furs is precisely 30% (which may or may not be a good assumption, see below).

To calculate the uncertainty, I used the fact that this is a binomial distribution. The formula for the variance of the number of yeti furs collected (which I'll call y) is therefore:

σ² = np(1-p)

where n is the number of adventures spent, and p is the drop rate expressed as a decimal. Since p is the number we are trying to find out, we don't know it for sure and approximate with the observed drop rate (r = y/n) instead. Since r = y/n, we can use the formula Var(nX) = n²Var(X) and see that the variance in the observed drop rate is:

σ² = r(1-r)/n

and therefore

σ = √(r(1-r)/n)

but note that this is the uncertainty in the drop rate, not in the actual +item drop bonus. Since the item drop bonus is shown by (yeti fur drop rate with blue tongue)/(yeti fur drop rate without blue tongue), the uncertainty scales in the same way:

σ = √(r(1-r)/n)/0.3

(0.3 = the 30% drop rate for yeti furs).

This formula means that if you want 10 times more accuracy, you need 100 times as many adventures spent on it. Ideally, I'd like 10,000 adventures' worth of data (which would lead to a σ of 1.6% in the item drop bonus.)

Once you have the value for σ note that there is a 95% probability of the actual +item bonus being within 2σ of the estimated +item bonus. This is where the plus-or-minus 10% comes from above.

If the uncertainty of the yeti fur drop rate is also taken into account, then things get even worse: if the drop rate is actually 31% instead of the 30% assumed, then the +item drop bonus estimate drops from 32% to 27%. --Rhebus 15:22, 16 February 2006 (Central Standard Time)

After even more testing between rhebus and me we now have the error down to 30.4+/-3.35% with 95% confidence. I think this shows pretty conclusivley its a 30% drop rate enhancer! --d0om 5 April 2006