Really Complex Travel Software Doesn't Exist For A ReasonMarch 15, 2015
I am trying to plan a trip to England and Germany to visit relatives I haven't seen since I was a kid and the planning is driving me nuts. Trying to coordinate flights and trains and get all the timing to work out is almost impossible.
My previous job was at a brand name online travel company (now sadly just a brand of someone else) so I know a lot of how the backend of travel works. But that's not any benefit. Travel is ridiculously complicated, and the lack of good integration and no easy way to make money makes building anything that would help me impossible.
Airlines pay at most $10 per booking regardless of complexity and this fact alone killed most of the travel agency business. Of course if your flight is simple booking online at an OTA or an airline website is easy. Try doing something more complicated and you are basically screwed.
I have to fly back and forth to London and within that stay fly back and forth to Germany plus either drive or take a train to a small town where my family came from. Complicating that fact is my size, as I need to find seats which can fit my lanky 2 meter (6'5") frame. Planes, trains and automobiles is a fine movie title but a nightmare planning scenario.
Most airline and OTA sites do not like multi-destination trips such as 3 sided (ABCA) or anything like mine (AB-CBA); these are simply not worth their effort. Building a simple flight itinerary (ABA) is hard enough given the enormous number of potential matches (as I've talked about in my flight articles, Part 1 and Part2 ); the more complex the pattern the more impossible to get any decent performance on building a list becomes.
I know there are travel agents who specialize in certain combinations who often know better deals than any online system can generate but they are not easy to find.
You would think with all the awesome computing power we have today we could solve this type of arbitrary reservation search but the issue is that most of today's reservation data is still maintained in 50 year old mainframe systems (mostly SABRE and Amadeus) which are pathetically incapable of anything more than what they were designed to do. Virtually all airlines still use these dinosaurs as their back end, as building a flight engine is a major pain and it's cheaper to just keep using them than to build a new one. About the only new edition is the engine behind Google Flight but it is not a full reservation engine (you can't book from their data) since virtually all airlines use the two dinosaurs.
Algorithmically finding a set of available, acceptable, legal and time correct flight choices between two or more airports is still a major problem to design, since there are so many variables to consider; add to that that prices are created with highly complex rules which makes it even more difficult.
Include the option of taking a train or driving or a ferry or other modes of transportation and have a system tell me the "best" way to make my trip work is likely impossible or at least economically infeasible to create. If you had access to up to date data all in one place it might be doable, but sadly the data is scattered (sometimes deliberately), broken and stale making it tough to imagine ever having such a system.
The trip might be fun if I can ever get all of this to work together. This is one startup you won't ever see no matter how much money you raise: plan a trip from anywhere to anywhere via anything.
I think planning a move to Mars is simpler. Maybe I should go there instead.