dcramer / peated

https://peated.com
Apache License 2.0
59 stars 12 forks source link

Reconsider Brands and Distillers #179

Open dcramer opened 3 weeks ago

dcramer commented 3 weeks ago

What do they do today?

Well simply put, they're label groupings.

What is the issue?

They get quite complicated and confusing, and this confusion leads to data entry issues.

For example, lets use the Mars brands from Japan. Mars owns a number of distilleries and bottles whisky under the Mars brand, as well as the distillry brands (e.g. Mars Tsunuki):

https://peated.com/bottles/13218

The Mars brand itself provides us very little value in this case, as "Mars Shinshu" would likely be a brand in our schema, but its actually just the distillery.

In other situations there are bottles like Mars The Y.A., which are bottled by Mars, and a blend of their distilleries:

https://peated.com/bottles/11829

What can we do about it?

Well lets consider what happens if we remove brand, and start there.

In some cases, Brand is actually already represented by Bottler. Even if its not, the value it adds is not very large. Most folks care about the distilleries at the end of the day anyways.

The primary issue comes up when we have blends from multiple distilleries. Some of these are cleanly handled via a Bottler such as SMWS. This ones an interested exercise because they are the bottler, but we also consider them the brand. Really we don't care about the Bottler at all and only want one value to understand who is producing the bottle vs where was the whisky distilled..

Another option when it comes to brands: remove the joint entity table. A big part of the issue is we have brands that are named X and then distilleries often named [Y] Distillery. These should be the same, but technically speaking calling them the same is often wrong. The reason we do this is because you don't want to duplicate data entry. For example, High West Distillery. The brand is High West (albeit they have several different labels...) and the Distillery is High West. In this world we could simply remove the Distillery suffix from all entities and ensure they're flagged correctly.

What do I think?

The safe path is to more cleanly merge distillers and brands, and start there. There's a number of things that are easily improved (search, merge of duplicates, name normalization) that might reduce the impact of this. We can also show the distilleries more front and center on the bottle details.

dcramer commented 1 week ago

I've mostly cleaned things up, and we now show distillers more prominenly. I'm ok with the current status, but will leave this active to track future thoughts.