graphs on the d3 screen are not using colors

[tzachi-dar]

We should have less than 80 as red and more than 180 as yellow.

[tzachi-dar]

Just sbmitted a pull request to this issue..

[bewest]

Howdy, it might be good to get more awareness around the design of nightscout, http://nightscout.github.io/posts/colors-subdued-is-safer/

We are interested in providing lots of freedom and choices, but at the same time, the black and white color choices were not made lazily or haphazardly.

[scottleibrand]

While I agree that muted colors should be the default for
cgm-remote-monitor, I do think colors should be an option for the
android-uploader, just as they are for cgm-remote-monitor. Colors are
better for understanding retrospective data at a glance. And in the case of
android-uploader, the display is not an always-visible, and isn't using
bright colors to display BG alerts like cgm-remote-monitor does, so using
colors there doesn't pose much risk IMO.

-Scott

On Mon, Nov 3, 2014 at 3:42 PM, Ben West [email protected] wrote:

Howdy, it might be good to get more awareness around the design of
nightscout, http://nightscout.github.io/posts/colors-subdued-is-safer/

We are interested in providing lots of freedom and choices, but at the
same time, the black and white color choices were not made lazily or
haphazardly.

—
Reply to this email directly or view it on GitHub
#56 (comment)
.

[tzachi-dar]

It's a nice post about the colors.
So, I guess that all lines should stay as they are while dots above the maximum, or below the minimum should be red and yellow.
This means that only in the case of something above 180 or below 80 there will be colors. Does this sound reasnable?

[bewest]

There are a lot of preferences in this area, so what I propose is that we create a few options to vary the behavior with colors, along with ways to measure at how effective they are over time.

Can we create a button or a tracker to measure effectiveness/safety of different approaches, and then use data to figure out which approaches might work best? Personally, I don't have a strong feeling for or against colors, but am more concerned with the process used to decide and implement and test these types of things. For example, I know that the design rationale behind limited color use and pop-out red was to assist in spotting situations where some kind of interference or action is required, and to make those situations uniquely differentiated from all the other times where people simply glance and look away. What is the design rationale for yellow and red, for example? Is it a potential "bike shed" scenario? Why not purple and cyan, for example? If we adopt red and yellow, can we measure whether or not they used, whether or not they perform better at situational awareness? Do they drag people into the display to look longer, or free people up to do more desirable activities?

[bewest]

What if the coloring was limited to the past half hour, and faded to gray/back/white?

[tzachi-dar]

This is probably even a better option.

I guess that we should be gray unless we are above the high bar or below
the yellow bar for the last half hour.

Once we have a reliable prediction that can also use colors (in the case of
danger).

So, do we have an agreement?

Thanks
Tzachi

On Tue, Nov 4, 2014 at 8:59 PM, Ben West [email protected] wrote:

What if the coloring was limited to the past half hour, and faded to
gray/back/white?

—
Reply to this email directly or view it on GitHub
#56 (comment)
.

[scottleibrand]

Whatever you do, it should start out as a non-default option that can be
enabled in preferences.

-Scott

On Tue, Nov 4, 2014 at 11:23 AM, tzachi-dar [email protected]
wrote:

This is probably even a better option.

I guess that we should be gray unless we are above the high bar or below
the yellow bar for the last half hour.

Once we have a reliable prediction that can also use colors (in the case
of
danger).

So, do we have an agreement?

Thanks
Tzachi

On Tue, Nov 4, 2014 at 8:59 PM, Ben West [email protected]
wrote:

What if the coloring was limited to the past half hour, and faded to
gray/back/white?

—
Reply to this email directly or view it on GitHub
<
https://github.com/nightscout/android-uploader/issues/56#issuecomment-61694368>

.

—
Reply to this email directly or view it on GitHub
#56 (comment)
.

[tzachi-dar]

Hi Ben,

you have many points there. I guess that red and yellow are well known
colors for dandger around the world.
In our scenario we can also use blinkings and alerts (I guess we should
actually use them)...

I'll think of the ideas that I have and that can be implmented in reasnoble
time and send it for review, so we can all be on the same line before we
start coding.

Tzachi

On Tue, Nov 4, 2014 at 8:58 PM, Ben West [email protected] wrote:

There are a lot of preferences in this area, so what I propose is that we
create a few options to vary the behavior with colors, along with ways to
measure at how effective they are over time.

Can we create a button or a tracker to measure effectiveness/safety of
different approaches, and then use data to figure out which approaches
might work best? Personally, I don't have a strong feeling for or against
colors, but am more concerned with the process used to decide and implement
and test these types of things. For example, I know that the design
rationale behind limited color use and pop-out red was to assist in
spotting situations where some kind of interference or action is required,
and to make those situations uniquely differentiated from all the other
times where people simply glance and look away. What is the design
rationale for yellow and red, for example? Is it a potential "bike shed"
scenario? Why not purple and cyan, for example? If we adopt red and yellow,
can we measure whether or not they used, whether or not they perform better
at situational awareness? Do they drag people into the display to look
longer, or free people up to do more desir able act ivities?

—
Reply to this email directly or view it on GitHub
#56 (comment)
.

[tzachi-dar]

OK, here are my thoughts:

As for colors in the case of emergency we might want to use background
colors as well. For example in the case of bg < 80 the entire screen can
become red... This can allow for a much richer use of colors, but with
higher coding effort.

So, there are the following color modes that I see:

1. Graph is all gray, (as today ?) the arrow + number are gray. In case of
   dangers they (arrow and value) become red or yellow. (Depending on value)

2. Like the dexcom receiver: one red line, one yellow line. Everything
   between them is gray, above the yellow is yellow, below the red is red. In
   case of dangers they (arrow and value) become red or yellow.

3. Like 2, but in the case of danger, the entire screen gets yellow/red.

More things besides colors that should be configurable in the UI:

1. Linear to log mode in the Y axis.
2. The y-axis lines can be configurable. not all people can live with 80 -

As a side note, I also want to fix the bug that the dots cover the numbers
on the Y axis as well :-)

Please let me know what you think.

Thanks
Tzachi

On Wed, Nov 5, 2014 at 12:39 AM, Tzachi Dar [email protected] wrote:

Hi Ben,

you have many points there. I guess that red and yellow are well known
colors for dandger around the world.
In our scenario we can also use blinkings and alerts (I guess we should
actually use them)...

I'll think of the ideas that I have and that can be implmented in
reasnoble time and send it for review, so we can all be on the same line
before we start coding.

Tzachi

On Tue, Nov 4, 2014 at 8:58 PM, Ben West [email protected] wrote:

There are a lot of preferences in this area, so what I propose is that we
create a few options to vary the behavior with colors, along with ways to
measure at how effective they are over time.

Can we create a button or a tracker to measure effectiveness/safety of
different approaches, and then use data to figure out which approaches
might work best? Personally, I don't have a strong feeling for or against
colors, but am more concerned with the process used to decide and implement
and test these types of things. For example, I know that the design
rationale behind limited color use and pop-out red was to assist in
spotting situations where some kind of interference or action is required,
and to make those situations uniquely differentiated from all the other
times where people simply glance and look away. What is the design
rationale for yellow and red, for example? Is it a potential "bike shed"
scenario? Why not purple and cyan, for example? If we adopt red and yellow,
can we measure whether or not they used, whether or not they perform better
at situational awareness? Do they drag people into the display to look
longer, or free people up to do more desir able act ivities?

—
Reply to this email directly or view it on GitHub
#56 (comment)
.