6 ms·

Why Good Visualization Matters: Rethinking the Food Label

41 points by apievangelist 15y ago

astarwithin 15y ago
I like knowing how much of things is in my food. :)
hugh3 15y ago
I'm not sure this is great visualization. Firstly, trace ingredients are often what's important to the consumer. Fine, I know my peanut butter is mostly peanuts, but how much salt is there? Secondly, these are simple examples. What do you do with, say, a can of chicken and vegetable and noodle soup, which could easily have thirty ingredients before you even start counting the potentially-nasty additives down in a tiny box in the corner. Thirdly, apparently labellers aren't obliged to split up certain categories (eg broccoli, sugar snap peas, green beans and carrots are all in a single green square). So presumably this doesn't even solve the specified problem of not knowing how much white vs whole wheat is in your food -- why would you split up white and whole wheat into separate categories when beans and carrots are the same?
- baconner 15y ago
  I didn't think the #1 entry was very practical either, but the full list of entries is worth a look. There are some good ideas and more realistically usable designs in there. GOOD just isnt likely imo to choose winners based on usefulness over beauty.
- calebmpeterson 15y ago
  > Secondly, these are simple examples. What do you do with, say, a can of chicken and vegetable and noodle soup, which could easily have thirty ingredients before you even start counting the potentially-nasty additives down in a tiny box in the corner. How about inverting the tree-map such that the smaller the amount, the greater the portion of the tree-map taken up? You're going to need all of that space anyway for the names of those potentially-nasty additives to be printed in a legible font size...
  - hugh3 15y ago
    Ah, the homeopathic approach to food labelling. I'm gonna sprinkle a grain of gold dust on every twinkie I sell, just to screw with the label.
- backprojection 15y ago
  > Firstly, trace ingredients are often what's important to the consumer. Fine, I know my peanut butter is mostly peanuts, but how much salt is there? You could put it on a log scale. Probably still confusing to people not familiar with logarithms.
notatoad 15y ago
it's a cool idea, but i believe most manufacturers would claim 'trade secrets' protections before putting proportions on their labels.
- jzb 15y ago
  They'd claim something to avoid having to make it easy to visualize just what's in processed food...
rorrr 15y ago
His example with 5 ingredients in unreadable already. What will happen with 10-20?
oscilloscope 15y ago
A few friends of mine built something similar. Loaded the USDA nutrition database in a dropdown, and scaled 24 vitamins, nutrients, calories, etc by recommended daily intake: http://exposedata.com/intake/foodpick.html http://exposedata.com/intake/foodpick.html (takes a minute to load) Many people have told me an added layer of discovering foods with particular nutritional qualities (filtering, sorting) would be very appealing. A recommendation engine for food. If you're looking to hack on the dataset, check out this JSON version: http://ashleyw.co.uk/project/food-nutrient-database http://ashleyw.co.uk/project/food-nutrient-database
- andreyon 15y ago
  wow, that's really cool! can they make a side by side comparator? it doesn't have to actually compare, can be just another frame ( or frames) where I can select the same and do the comparison myself. For example I want to see the differences in raw broccoli and steamed broccoli.
  - oscilloscope 15y ago
    Open two browser windows ;) Here are a few other concepts we messed around with: http://fleetinbeing.net/intake/web/rdi.html http://fleetinbeing.net/intake/web/rdi.html (stacked bars) http://fleetinbeing.net/intake/web/groups.html http://fleetinbeing.net/intake/web/groups.html (hover to see vegetable/nutrients) http://fleetinbeing.net/intake/web/matrix.html http://fleetinbeing.net/intake/web/matrix.html (matrix plot)
aaronblohowiak 15y ago
IMHO, the Nutrition Information box is actually a shining example of good design already. It is very, very legible and has a clear information hierarchy.
rossriley 15y ago
The new food labelling in the UK is based on a traffic light principle and, imho seems to be clearer: http://newsimg.bbc.co.uk/media/images/44434000/jpg/_44434134_foodlabel203pa.jpg http://newsimg.bbc.co.uk/media/images/44434000/jpg/_44434134... The problem with this exaple is that it's hard to see quickly what the key points are, since prominence is automatically given to higher scores, whether they're good or bad.
SeanLuke 15y ago
This is pretty bad I think. The amount of substance by mass is nearly completely unimportant. Consider a package which has 99.999% Organic Apples and 0.001% Plutonium. But hey, it's 99.999% organic! Look at all that red! Much better is a simple ordered list of ingredients, so we're not bamboozled into ignoring certain ingredients just because they're not the majority of the content. But wait, thats ... what we already have on our packaging? What good visualization!

41 points by apievangelist 15y ago

I like knowing how much of things is in my food. :)

hugh3 15y ago

I'm not sure this is great visualization. Firstly, trace ingredients are often what's important to the consumer. Fine, I know my peanut butter is mostly peanuts, but how much salt is there? Secondly, these are simple examples. What do you do with, say, a can of chicken and vegetable and noodle soup, which could easily have thirty ingredients before you even start counting the potentially-nasty additives down in a tiny box in the corner. Thirdly, apparently labellers aren't obliged to split up certain categories (eg broccoli, sugar snap peas, green beans and carrots are all in a single green square). So presumably this doesn't even solve the specified problem of not knowing how much white vs whole wheat is in your food -- why would you split up white and whole wheat into separate categories when beans and carrots are the same?

baconner 15y ago
I didn't think the #1 entry was very practical either, but the full list of entries is worth a look. There are some good ideas and more realistically usable designs in there. GOOD just isnt likely imo to choose winners based on usefulness over beauty.
calebmpeterson 15y ago
> Secondly, these are simple examples. What do you do with, say, a can of chicken and vegetable and noodle soup, which could easily have thirty ingredients before you even start counting the potentially-nasty additives down in a tiny box in the corner. How about inverting the tree-map such that the smaller the amount, the greater the portion of the tree-map taken up? You're going to need all of that space anyway for the names of those potentially-nasty additives to be printed in a legible font size...
- hugh3 15y ago
  Ah, the homeopathic approach to food labelling. I'm gonna sprinkle a grain of gold dust on every twinkie I sell, just to screw with the label.
backprojection 15y ago
> Firstly, trace ingredients are often what's important to the consumer. Fine, I know my peanut butter is mostly peanuts, but how much salt is there? You could put it on a log scale. Probably still confusing to people not familiar with logarithms.

notatoad 15y ago

it's a cool idea, but i believe most manufacturers would claim 'trade secrets' protections before putting proportions on their labels.

jzb 15y ago
They'd claim something to avoid having to make it easy to visualize just what's in processed food...

rorrr 15y ago

His example with 5 ingredients in unreadable already. What will happen with 10-20?

oscilloscope 15y ago

A few friends of mine built something similar. Loaded the USDA nutrition database in a dropdown, and scaled 24 vitamins, nutrients, calories, etc by recommended daily intake: http://exposedata.com/intake/foodpick.html http://exposedata.com/intake/foodpick.html (takes a minute to load) Many people have told me an added layer of discovering foods with particular nutritional qualities (filtering, sorting) would be very appealing. A recommendation engine for food. If you're looking to hack on the dataset, check out this JSON version: http://ashleyw.co.uk/project/food-nutrient-database http://ashleyw.co.uk/project/food-nutrient-database

andreyon 15y ago
wow, that's really cool! can they make a side by side comparator? it doesn't have to actually compare, can be just another frame ( or frames) where I can select the same and do the comparison myself. For example I want to see the differences in raw broccoli and steamed broccoli.
- oscilloscope 15y ago
  Open two browser windows ;) Here are a few other concepts we messed around with: http://fleetinbeing.net/intake/web/rdi.html http://fleetinbeing.net/intake/web/rdi.html (stacked bars) http://fleetinbeing.net/intake/web/groups.html http://fleetinbeing.net/intake/web/groups.html (hover to see vegetable/nutrients) http://fleetinbeing.net/intake/web/matrix.html http://fleetinbeing.net/intake/web/matrix.html (matrix plot)

aaronblohowiak 15y ago

IMHO, the Nutrition Information box is actually a shining example of good design already. It is very, very legible and has a clear information hierarchy.

rossriley 15y ago

The new food labelling in the UK is based on a traffic light principle and, imho seems to be clearer: http://newsimg.bbc.co.uk/media/images/44434000/jpg/_44434134_foodlabel203pa.jpg http://newsimg.bbc.co.uk/media/images/44434000/jpg/_44434134... The problem with this exaple is that it's hard to see quickly what the key points are, since prominence is automatically given to higher scores, whether they're good or bad.

SeanLuke 15y ago

This is pretty bad I think. The amount of substance by mass is nearly completely unimportant. Consider a package which has 99.999% Organic Apples and 0.001% Plutonium. But hey, it's 99.999% organic! Look at all that red! Much better is a simple ordered list of ingredients, so we're not bamboozled into ignoring certain ingredients just because they're not the majority of the content. But wait, thats ... what we already have on our packaging? What good visualization!