Idle Time » Blog Archive » Let users help solve their own problems

I wish we had a maps app like Apple Maps or Google Maps that would allow you to order a travel itinerary using public transit between two points, and clearly choose the travel routes that involved. Or, preferably, multiple sets of routes, for comparison.
Like, let’s say I’m on a Mission and I want to get to the Metreon. One of the existing apps may suggest the 14R rapid bus, which arrives in X minutes and takes Y minutes, or BART, which arrives in Z minutes and takes W minutes. But that leaves alternatives like taking the 49 bus to Van Ness Station and then taking the Muni subway from there.
Sometimes all of the app’s recommendations make sense, but sometimes there are one or more options that could be better—and I don’t know how it would be better if the app didn’t show me when the next 49 is coming, so I can compare it to 7 minutes for the 14R or 9 minutes (including a short walk) for BART.
What I would like is the ability to add a specific set of routes to be taken into consideration, or even force an itinerary using those routes. Let me say “49, KLMN” and include this series of routes in my options for comparison.
This is an example of a general problem, which is products that only have an algorithmic solution to the user’s needs, without the opportunity for the user to contribute to the solution.
The algorithmic-only model admits only one remedy: Improve the algorithm. But since no algorithm can be perfect, you’ll be playing this whac-a-mole game forever.
This is for the developer of each product, as well as the user. I can try the same question on Apple Maps and Google Maps and on Transit app and what else, but as long as they are all work this way, all I do is hold different slices of Swiss cheese next to each other and compare their holes.
If the user can contribute to the solution, then there is a chance that they will have a better idea of how to address their own needs.
And they are not equal to each other. You can use user input to improve the algorithm’s suggestions.
Email spam filters have had this for decades. The “Report Spam” and “Not Spam” buttons help train the filter. And we still have them because the filter will never be perfect (not only because spam is always evolving).
For the transit routing example, this is a more complex problem (not a simple ham-or-spam dichotomy) and there are privacy considerations. However, helping to develop a routing algorithm for everyone can be something people choose.
If they refuse, perhaps the algorithm can use training supplements stored locally. The user providing the suggestion can still benefit, even if they refuse to share that (possibly personal/identifying) data with others.
And even the ability to add a route combination to the list, although I have to do it every time, is an improvement without that and limited to whatever option the algorithm chooses for me.
Don’t think that your algorithm has to solve everything by itself. Help the user. They are happier with a solution they helped create, partly because it will be better for their specific needs, and partly because they need to be involved.
2025-01-15 00:33:00