I sat in the car park of a rental office for twenty five minutes with the keys in my hand, not driving anywhere, running the numbers again on the back of the contract. I had just committed to four days of a vehicle I was not sure I needed, in a region where I had already been told the roads were bad in a way the person telling me clearly thought was funny.
Two days later I was reversing three hundred metres down a track because it had narrowed to something my hire agreement did not cover, and a man on a motorbike was waiting patiently behind me, watching, offering nothing.
The maths I used to do wrong
My old method was to compare the daily rate against the cost of tickets, notice the tickets were cheaper, and take the bus. That comparison is almost meaningless and I used it for years.
It leaves out fuel, which in remote areas can be a serious number, and it leaves out the fact that a car is not one seat, it is unlimited seats within one trip. It also leaves out the cost of the nights I have to sleep somewhere I did not want to sleep because the connection only runs in the morning. Three extra nights of accommodation to make a bus itinerary work will eat a hire car whole.
The other thing it leaves out is what I actually get to see. A bus itinerary in a thin network is not a route, it is a series of forced stopovers in whichever settlements happen to be junctions. Sometimes those are the best part of the trip. Sometimes they are a petrol station and a closed restaurant.
What the car actually buys
The car buys the ability to change my mind. That is the whole product, and it is worth a lot in some places and nothing in others.
If I am going to one valley and staying in it, changing my mind has almost no value, and the car sits outside a building for four days depreciating my budget. If I am moving through a region where the interesting things are twenty kilometres apart in different directions with no connecting service, the ability to change my mind is the difference between seeing four things and seeing one.
It also buys light. Public transport in thin networks is a daytime activity, and it collapses in the late afternoon. A car keeps working after dark, which extends the usable day by hours, and those hours are often when I actually want to be somewhere.
What the bus buys that the car cannot
The bus puts me inside the local movement of the day. That sounds soft and I resisted saying it for a long time because it sounds like the kind of thing people say to justify a slow choice. But it is concrete.
On a rural service I find out where people are actually going, which is usually not where visitors go. I have gotten off at stops I had never heard of because everyone else got off there and it seemed like information. Twice that has been the best decision of a trip.
The bus also removes an entire category of stress: parking, fuel range, the state of the road surface, whether the track I can see on the map is a road or a memory of one. I am bad at estimating road quality from a map and I have consistently overestimated what a small hire car will do.
And the driver knows things. A driver who runs the same route daily is the single best informed person about that corridor, and on a quiet run they will tell me most of it.
The roads decide more than I do
The factor I underweight most is surface. Not distance, not the map, the surface.
Distances that look small collapse into whole days when the road is unsealed and rutted, and a small car with no clearance turns a two hour drive into a four hour crawl with a knot in my stomach. Meanwhile the local minibus does it in the advertised time because the driver knows every bad section by feel and does not care about the underside of the vehicle.
So now I ask a specific question before booking anything: is the road sealed the whole way. If the answer involves any hesitation, I assume it is not, and I assume it is worse than the hesitation implied. That one question has changed more of my decisions than everything else on this list.
How I choose now
I ask four things in order. Is there a service at all, running on the days I need. If yes, how many per day, because one per day is not a network, it is a commitment. If it is one per day, does the itinerary force me into stopovers, and are those stopovers places I would want to be. If the answer to that last one is no, the car starts making sense even at a worse headline price.
Then I check whether I am comfortable driving there. Not whether I can. Whether I want to spend the trip doing it. Some regions I would rather look out of a window in than concentrate on a road, and I have stopped pretending that is a failure of nerve. It is a preference, and it is allowed to be an input.
The trip I got backwards
The reversing incident was the car being wrong. I had hired it for flexibility in a region where the only things I wanted to reach were on a single valley road that had a twice daily service running exactly along it. I paid for four days of independence I never used, and spent one afternoon frightened of a track.
The opposite happened later that year. I tried to do a scattered region by public transport out of stubbornness, and lost two full days to waiting, plus two nights in a junction town that had nothing except somewhere to wait in. The car would have cost less than those nights and given me back the days.
Neither mistake was about money. Both were about applying a rule instead of looking at the shape of the place. Sparse and scattered wants a car. Linear and served wants the bus. I get it right most of the time now, and when I get it wrong I try to notice which of the two mistakes I made, because they are opposite and I am capable of both.