Hacker News Reader

Comment by 4b11b4 | Hacker News Reader

I've come back to this idea of "differences" from many angles...

From images: information _is _ the differences between pixels

From opinions: my opinion alone is worthless, but a difference of opinions? now we've got something

From relationships, graphs, networks: edges are _differences_ between nodes. the edges are everything. nodes alone... are meaningless

From music: this one is a little easier to take at face value, you can't have a sound without a change (difference) over time. the speaker cone can only wobble...

Not sure about the absoluteness or correctness of this intuition, please criticize

View on HN

For images, you need a reference pixel to get the color right. You can apply differences from there but you’re effectively working with absolutes because of that so it’s a meaningless distinction.

Any time an absolute reference point is assumed, all downstream calculations are effectively absolutes

Is this strictly true?

Suppose you have a reference pixel but don’t know its value. You don’t get the color right but there is plenty of other information there, right? Which pixel also probably limits the range of values since your highest and lowest values are probably within a certain range.

I have a blue/black (or possibly white/gold) dress that says it's relative.

You can also ask a UI designer about "absolute" colors. One color needs to be different depending on the context it's in order to LOOK the same (it's size/thickness, background, what it's next to).

The blue black dress thing is how people perceive it, but the values for pixels are absolute. The pixel values aren’t changing from viewer to viewer.

If OP had said “perceived color”, I would agree. But we’re talking at pixel level here.

For music, a note is just a frequency. It is the difference between the notes that makes music, not the notes themselves

"Information: any difference that makes a difference" - Gregory Bateson

Reply on news.ycombinator.com