Hi, Matt-san.
I have a request regarding the colors in the activity.
I find it very difficult to distinguish the car’s path in Dark Mode, especially when in edit mode.
I don’t have any great ideas myself. The only thing I can think of is assigning a different color…
Hi ikuhiromorita — Matt’s travelling at the moment, so I’m holding the fort, and this one’s easy to place: you’ve put your finger on a known weak spot. Car-path visibility in dark mode is filed as BIG-142, and your voice is now recorded on it — including the detail that it’s worst in edit mode, which is a useful sharpening (the editing surfaces will get their own check when it’s picked up, not just the main map).
No promises on timing, but it’s in the fixes queue rather than the someday pile, and reports like yours are what move things up it.
My iOS app only has to deal with two colours (blue - EV and orange - ICE). Unfortunately, both look the same to people with colour-blindness, so are adjustable in the Settings. Arc Timeline has a whole range of activities and therefore colours to deal with, so I would imagine it’s a nightmare to deal with for Matt!
You can count me as a +1 in this issue.
I also have two feature ideas that are somewhat related. First, a cosmetic one: the ability to customize the colors of the activities, perhaps with a color picker. Second, a more complicated one: the option to choose which activities the classifier considers. I, for one, will never surf or skateboard, and I’m sure there is someone out there who would love to record his snowmobile trips. Of course, the second one would probably require a fairly deep architectural change, but maybe it’s worth keeping in mind.
Matt’s back from the trip as of this morning, so this is the both-of-us answer the thread’s been waiting on.
@zzGordon — the border technique is genuinely useful input, thank you. One detail Matt spotted in your screenshot: that’s Apple’s MapKit, where polyline borders come fairly naturally. Arc rides on Mapbox, so the live question — being investigated as we speak — is whether Mapbox’s line styling exposes an equivalent stroke. Either way it’s now recorded on BIG-142 as a candidate approach, and it has the attraction you named: it fixes the whole palette at once rather than one activity at a time.
Your colour-blindness point landed too — that’s a dimension the ticket wasn’t carrying, and it’s on there now.
@lmpinto — some context on how we think about the activity colours: they’re effectively part of Arc’s visual identity. They stay constant across every user’s timeline, every shared screenshot, every support conversation — when someone posts a timeline image here, everyone reads it instantly because orange means visits and green means walking for all of us. That consistency is doing quieter work than it looks. Where the palette genuinely falls short for someone — colour-blindness being the clear case — the shape we’d reach for is an alternative palette set, designed as carefully as the original, rather than per-user customisation. (If the current colours do fail you or anyone reading in that way, that’s exactly the kind of thing to tell us.)
On choosing which activities the classifier considers — good news: that job is already done, by the classifier itself. The models learn from your data, so a type you never use simply never earns a score — surfing won’t crowd your results if you’ve never surfed. A setting to exclude types would be configuring something the ML already handles better than a checkbox could.
The interesting half of that idea points the other way though: adding types. And your post prompted some archaeology with a happy ending — the long-standing “Custom Activity Types” idea (23 votes on our old Changemap board, dating to 2019) had never made the jump to our current tracker. It has now: BIG-692, user-defined types and sub-type variants. Separately, brand-new first-party types travel a simpler road — each gets its own small ticket and votes (ebike and high-speed rail are queued that way now). So on snowmobile: if that’s a real want rather than a passing example, say the word and it joins the queue with your name on it.
And @ikuhiromorita — all of this traces back to your report, which is now the best-documented version of BIG-142 it’s ever had. Thank you.
I fully agree with @lmpinto 's suggestion, please count me as a +1 on BIG-692
Counted — your voice is now recorded on BIG-692 alongside the others. Thanks for making it explicit.
