A demo app being created by MikeCa, with help by TonyLam and DanFr.

The SodaApp is a filtering application where all users of a group have their Soda ratings and consumptions brought together to form a list of what should be purchased. See SodaApp Introduction for how to become part of the consuming nation. Go to the @Soda page if there's a Soda you don't see and you want to add it.

More notes from the soda buyer, 6/2.

• I notice that UdVenkat still has votes logged, even though presumably he won't be drinking too much GroupLens soda any more. This is probably a general problem. What's the right way to deal with it?

DanFr: You know, I just went to his page, and edited it to 0, and I like that. DanCo wasn't so comfortable with Mike's editing of his page and so didn't want to go poking at others' pages.

• If people want to re-specify soda prefs, including including sodas that you like but aren't there, that'd be good. I'll probably look to buy another dose around the end of the month in order to balance our current inventory toward what people like.
• Sue bought a bunch of stuff based on her understanding of GroupLens, which revolves around MountainDew. She also got a bunch of deeply discounted water.
• I'm looking to turn over my stewardship of soda soon. Any takers?

FYI, I copied the state of the SodaApp on 4/22, 1:30 PM, and will use it as input to the soda-buying process.

Here's a quantized soda plan based on what I saw.

• 1 Case: Coke, Diet Coke, Water, Minute Maid Lemonade
• 1/2 Cs: Cherry something (Pepsi preferred), Dr Pepper, Mountain Dew, Pepsi
• 6-pack: Orangina, Diet Dr. Pepper.
• Ugh: Sprite, and various non-sodas and also-rans.

Here's what I "learned" that I didn't already know:

• Minute Maid Lemonade is much more popular than I believed.
• Sprite, even less popular than I believed.
• In general, there's a mild to moderate Coke company preference.

Note that "learned" is in "quotes" because it's only true if everyone went between Loren's announcement on Monday and my cutoff this afternoon. In particular, it might be that some people never rated sodas that were added later in the process.

What I do with this info I'm not positive, because I'm still constrained by trying to buy soda Real Cheap, but I will definitely lean Coke and be sure to get a few more non-carbonated things (including the Minute Maid). I probably will buy to supplement the current inventory (i.e., no Diet Dr. Pepper or Root Beer since we have them), will buy as things go on sale, and will try to buy 1-2 months' supply.

DanCo feedback: Cool idea, looks fun. Not knowing how it actually computed soda preferences from total soda and ratings left me feeling a bit in the dark but once I saw it it worked fine. I had a little trouble getting a working TotalSoda line; I had to cut and paste from someone else's, not sure why. The app should probably do some robust checking (e.g., over-large soda consumption, long names, etc.) but I suppose prototypes aren't expected to be perfect. Doesn't deal with substitutability well (if there is any soda that I like, it'd be okay, and that could simplify the buying process), but that could be tough.

A couple of formatting comments: 1) move the totals to the left and top. 2) Sort by quantity by default, rather than name, both vertically and horizontally. Sorting options would be nice, but work; the most likely best ordering is one that brings out salient facts, which for this app is all about quantity.

DanFr: Substitutability (a mouthful!) is a good issue. I actually gave this some thought, but it hasn't been implemented yet. For example, suppose the first round produces the answers above, the for all sodas that have fewer than N cans (N=12?), it eliminates those sodas from the running entirely, then runs again on the remaining? This would be a version of substitution: you can't have that very unpopular soda, but then I'll reapportion your budget among the remaining.