This post is related also to the other post about "Standard Format for truncated ethereum address"
Relaying here what the discussion was in Telegram
Would it be possible to get a job story or problem statement for the context when needing to differentiate between these things? I have my assumptions, but also misgivings about a generated visual indicator
As a ____ I need to _ so that I can __
[11:48:38 AM] Sarah Mills:
Yeah, I think defining the reason/job story would really help--there would be a difference in solutions between "I need to be able to quickly identify" vs. "I need to match this to..." etc. I suspect there is a difference between devs and novice user needs as well
[11:56:54 AM] Beltran
At ENS we are making the assumption that the Identicon job story is
"As a user, who inputs a hash in an input field, I need to verify that it is the correct hash, so that I can know I didn't make mistakes"
The identicon is a control
the idea is that you see the official links with their identicon
and when you paste it into an input field you can verify that it's the same
it's used to copy paste ( or automatically assigned) the default registrar and resolvers.
also another point where identicons are used is to quickly identify which domains you are the owner of and which you are not, because you have transferred them
[11:57:38 AM] Alejandro Machado
the key question is: are users going to be using the same address often? is the address they need to recognize something they've just seen somewhere else, or something they've seen many times in the past?
[11:58:37 AM] Pedro Gomes:
I think so, it’s probably one of the biggest differences between Bitcoin and Ethereum
Where a public address feels more personal
[11:58:53 AM] Sarah Mills:
Yeah, I was gonna say—I would love to see that tested, if people can indeed use that image to verify something is the same
[12:01:06 PM] Alejandro Machado
I'd love to see the efficacy of the three different identicons tested
[12:01:17 PM] Sarah Mills:
This would be such a good problem for a hackathon where design was included
because unless you have a list somewhere to compare the visual against (like your "contacts") I'm not sure how to test this
[12:03:06 PM] Aqeel Mohammad
just to make note the 3rd spot color on the ethereum blockies adds a lot to it.
[12:03:06 PM] Sarah Mills:
and then we run into anonymity issues
the low-hanging fruit use case for identicons is just making transactions to addresses you've just seen. e.g. if Shapeshift used identicons, I'd be a lot more reassured that I'm making a transaction to the right address, even if I just copied and pasted it
has anyone ever tried to make identicons that are also machine-readable? (another way to look at it, QR codes that are human-readable)
[12:07:30 PM] Pedro Gomes:
I think this would make a great online game
Where people could test themselves on memory sort of thing
And share the results with friends on social media
That would scale the experiment
[12:08:41 PM] Sarah Mills:
last thing I will say before continuing later on conflux—whatever the context and solution, I would love for there to be a non-visual indicator as well for people with low or impaired vision
[12:09:53 PM] Aqeel Mohammad
@alemac HRQR is a thing, but normal QR codes can have human readable or visual elements embedded within them.
[12:27:08 PM] Alex Van de Sande:
“You know what’s also machine readable? Helvetica”
Seriously, before you go in a rabbit hole of making QR codes more human readable, remember that it’s relatively easy for computers to read good type, and it has about the same information density
[12:27:57 PM] Alejandro Machado
oh yeah, but if we're going to truncate hashes they're not gonna be very readable
good point though
[1:54:19 PM] Alex Van de Sande:
I’m thinking that for things like private keys, word seeds and urls, just plaintext (maybe with a special dotted border so the computer knows where to scan) work as well as a qr code, are prettier and human readable
> @SarahMills: "Yeah, I was gonna say—I would love to see that tested, if people can indeed use that image to verify something is the same"
I would imagine two good tests for Identicons: rememberability (is that a word?) and comparability.
For the first one, I’d show 9 identicons for 30 seconds and then show a new set of dozens of icons and ask the user to select those he had seen on the previous screen. The second test would be a just some sort of how quick you can tell if this is 2 icons are the same or different
One last word on ethereum blockies: not only they have 3 colors (third one rarer so it forms dots for “eyes” and “noses”) but also the latest version has an option to make the shapes less “blocky” with circles and diagonals. So if you’re testing comparing them, make sure you include these.
I’d also be curious on testing things like “robohash” that create actual robot/monster faces (although they don’t have a lot of entropy)
[2:09:05 PM] Alejandro Machado
I think the word you're looking for is "recognition" (contrast to recall). I'd love this test
more info https://www.nngroup.com/articles/recognition-and-recall/
Nielsen Norman Group
Memory Recognition and Recall in User Interfaces
Showing users things they can recognize improves usability over needing to recall items from scratch because the extra context helps users retrieve information from memory.