mixnjuice / project-ideas

This is an attempt to create the best, most user-friendly mixing website the world has ever seen :)
9 stars 2 forks source link

Rating recipes and how to reduce the noise / fake reviews? #13

Open pscn opened 5 years ago

pscn commented 5 years ago

It shouldn’t look outdated or cluttered as ID10-T wrote, and it shouldn’t let people add nonsense flavours and fake reviews like it’s possible on ELR.

I think the first parts are a given, but I'm not to sure how to handle the last part.

To my understanding ATF has the subscription model in parts to reduce the noise of "fake reviews". Aka somebody just giving a recipe five stars without even mixing it. On the other hand ATF, in my opinion, has the problem of "too little" reviews because rating a recipe means too much effort. I'd like to be able to rate a recipe quickly, without having to enter any text or give it any more thought than "i like it" or "i don't like it".

We could have the following scenarios for rating and maybe weight each type differently when calculating the overall rating for a recipe:

For weighting the rating I'm thinking about: 0.1 for a simple, 0.2 for an advanced and 1.0 for a detailed review. Maybe also adjust the weight for the detailed review based on how much users found it useful / not useful.

Korlimann commented 5 years ago

I like the way this is built up. Especially the idea with badges or some kind of achievement you get when you rated a certain amount of recipes. We could maybe even provide some motivation to get them, say for example a user that has rated like 100 recipes has some "extras", that are not that big of a deal, something that's "nice to have" but not needed. One thing I just thought off: We could set something up so users can vote for new features. The people that contributed most to the website have a second vote, maybe even a third. So if you want to be able to decide which features should implemented next, you'll definitely want to be able to have more influence on that.

The rating itself seems good to me, although there is a huge gap between an advanced and detailed review. Maybe make it 0.2 for a simple, 0.5 for an advanced and 1.0 for a detailed review?

pscn commented 5 years ago

Voting for features is a nice idea.

We should collect ideas for badges somewhere to keep track of them. But always keep them tied in with other users feedback and make that feedback hard to fake. E. g. just having a badge for publishing recipes might give the wrong incentive. But giving out a badge for recipes with more than 10 mixes, / reviews etc. might be a better idea.

I'm not dialed in with any value for weighing ratings (or anything) right now. Just putting out ideas to have a basis for discussion. I'd expect values like that to be configurable somewhere so they can be fine tuned later.

Korlimann commented 5 years ago

I'd suggest that some badges are needed to have more votes, while others are just for shits and giggles. I'll create an issue for that rn

And yeah, you're right on the ratings, that's probably a thing that we'll tune later when everything is actually up and running

nickdurcholz commented 5 years ago

Regarding the trash data, I think everyone can agree that ELR is a mess, and well intentioned users cajoling others into using standard flavor names is not a solution. However, users don't always know what they've got or have some weird thing that isn't on the official list. I'm thinking of things like CBD, stevia products from vendors that don't market it as a vape product, home mixed additives like citric acid, and flavors from uncommon sellers who rebrand common flavorings and don't tell you who the manufacturer is. I made the mistake of buying from one such uncommon seller when I first started mixing, and I still have a couple bottles where I'm not really sure if what I have is Black Cherry from TPA, FW, FA, or a one-shot of random cherry flavors.

The way you would use such an ingredient on ATF is to use a well-known flavor as a stand-in for your special ingredient, which skews stats and is lying to others if they make the recipe public. On ELR, where it is entirely uncurated, the situation is worse because they have dozens or a hundred misspellings and typos of well-known flavors as well as flavors such as "!!! PISS !!!" and "dog shit".

If you want to support these people's use case, and we plan support for private recipes, maybe we could allow it in private recipes but prevent them from making anything that uses unknown ingredients public. We would probably also need to exclude recipes that include them from stats and anything else that could possibly affect others.

Of course, saying that we don't care about this use case is also an option. ATF has made that choice, and it doesn't seem to be suffering much from the decision, but I feel like a recipe app shouldn't be dictating what you are allowed to use, and it could cut down on the number of requests to add odd-ball flavors to the official list. BTW, how is THAT going to work?

pscn commented 5 years ago

Spot on. In #18 we were looking for a place to store the one-shots mixed with the app. I'm also with you on not allowing to publish recipes with these ingredients.

To make this more useful it would be nice to have the ability to change one of those "private" flavors with an official one, so it updates all your recipes accordingly. So if you have one obscure flavor and you're not 100% sure what it is, you can already add it to your stash and start mixing with it. Then later if you figured out what it really is, change it to the real deal.

JosefBud commented 5 years ago

On ELR, where it is entirely uncurated, the situation is worse because they have dozens or a hundred misspellings and typos of well-known flavors as well as flavors such as "!!! PISS !!!" and "dog shit".

Maybe it would be possible to try to match what they're putting in with something already in the database? If someone enters "TPA Bnana Crem" maybe try to match it and ask them if they actually meant TFA Banana Cream.