Triptales 0x0002: The Tools

Genetic algorithms used to be a popular optimization technique back in the '90s. And in a pathetic yet persistent attempt to name drop, I feel compelled to share that I got an 'A' in a graduate course by the person who pretty much mainstreamed the field. So why did I show off my nerdiness here? Sit down, young child... and let me share my wisdom with you. And if you are not familiar with the traveling salesperson problem, do read up on it.

I want to visit the 48 contiguous states. I want to do it in the shortest distance/time that allows me to cover the places I want to visit. I really don't want to retrace my route (looking at you, Nebraska and Kansas). The US being the 4th largest country in the world by area doesn't help the situation. Neither do the states of healthcare and employee benefits in the country (need my insurance, therefore need my job, and therefore need to minimize vacation time) help. The extremes of climate also play a role, e.g., cold winters in the north, brain-melting heat in the south. These are some "hard" constraints -- you can't really do much about them and you have to prioritize them.

Then there are the soft constraints. The Northeast part of the US has lovely fall colors as the foliage sheds. It'd be great to be there at that time -- around September. The weather gets colder up north later in the year -- it'd be nice to finish it up before the snowstorms begin. Some friends are not around at certain times. Work schedule might require me to be in certain areas at certain times. You get the drift, don't you?

So, I started with this.

A 3'x4' map clipped to a magnetic whiteboard with repositionable route markers and magnets was the "prototyping" phase. I was able to visualize the general route in my head so I could figure out where the issues were likely to be. Damn, Texas is big! As is California. Why is North Dakota so far away from anything? What am I supposed to visit in Idaho?

Remember how I started off this post with genetic algorithms? An enterprising soul (Dr. Randal Olson) has actually used genetic algorithms and created a way for you to enter a list of places you want to visit, and the program will then spit out an optimized route for you. There is some prework needed, of course. Me: I love planning for a trip. I use spreadsheets and charts.

Dr. Olson: Hold my beer.

Turns out Google Maps is not as useful in planning complex routes with lots of intermediate points either. It's hard to save a map/route and you can only add up to 10 intermediate waypoints in a route. You can work around by creating custom maps like this but then you lose the fancy routing.

So, after much handwringing and reading reviews, I finally gave in and bought a paid subscription to roadtrippers.com that allowed me to enter the list of places I want to hit and let the app do some basic routing. I believe it is value for money, and I haven't even started my drive yet.

But that still didn't quite cut it. The distance and time estimates were not as good as those from Google Maps. And the app was also trying to be too smart for its own good. It will modify the route as you add places to the route. Sometimes you have to go back and force it back to the original or your desired route. And you may have to artificially add locations so that the route "sticks" and doesn't get changed.

Here's the thing though. Visuals are great. But there are times I also need a table to filter, view, track stuff like confirmation numbers and addresses. Especially when you are on the road and can't be scrolling forever on a tiny phone to find some info.

Enter Google Sheets. The next step was to combine the information from the Roadtrippers app and Google Maps and create a spreadsheet that will allow you to see a consolidated view of where you are going to be on a certain day. And add whatever other information you want to. Here is a screenshot of my spreadsheet now.

This allows me to see at a glance where I expect to be on a certain day, what activities are planned for that day, whether it's a working day (or if I need to call in sic...vacation). The contact info for potential hotels/campsites are also there. I am tracking progress estimates, which are more for planning and making sure my mileage tracks the day count. In my first version, I was on day 90 and had just about shown up near Chicago -- I wasn't "pacing" myself well.

Using Google Sheets also means that I can share the sheet with family/friends and those who are standing by to call emergency services as needed. Oh, did I tell you that I am doing this mostly solo? The idea is that the kids and my join me for specific legs of the journey. This will also help figure out where they want to join and the nearest airport to fly in and out of.

A few friends have expressed interest in joining me for part of the trip. I intend to share this with them too so they can also plan things.

Putting it in a table also had the advantage that I could do rudimentary analytics on it. Like this.

The chart on the left shows estimated daily mileage. I look at the 300+ mile days and wonder what I can do to reduce them or make sure the 300- and 400-mile days are followed by easy days. Also, why is my route showing me spending 4 days in Salt Lake City? Why am I spending 2 days in Oklahoma City? Using analytics on data is like going to a therapist for counseling -- it's not about creating new stuff; it's about gaining new insights into data that you already have.

Post credit scene: Don't let Yoda distract you. Focus on the steering wheel.