Most of the travel industry runs on old software that would horrify a lot of people here, especially those who've never worked for a large, 30+ year old company. When I used to interview a lot of people I made it a point to mention some of the more "interesting" aspects so they'd know what they were getting into.
One example: ever tried to book a flight a year in advance? On a lot (almost all?) of systems you can't, because the underlying date format is "DEC27".
Edit to address a couple comments: logistics are hard and there are plenty of reasons why airlines wouldn't want to support booking that far out. However, the reason you can book a flight 330 days from now but not 360 days from now is almost certainly due to the date format. (I believe the windows used are less than 365 days because it's helpful to be able to have dates in the recent past. I remember seeing documentation for 360, but AA and United seem to be in the 330-340 range on their websites).
As a fun side thing, I am also a travel agent with access to some of these internal systems on the booking side. The technology is incredibly antiquated. Most of the US runs on a system called SABRE, which is basically a MS-DOS system with a text command line interface and its own language. It's all ASCII text based (and all in uppercase). It's straight out of the 80s. Travel agents need to buy special "errors" insurance to cover any losses caused by fucking it up (a typo could accidentally cancel a ticket and cause the client thousands in losses rebooking it).
They actually have a GUI interface over it now with the ability for power/legacy users to drop into the raw shell, if they wish. From feedback, many of the older agents actually prefer the command line, because it’s muscle memory and an experienced agent can perform routine tasks that would take multiple screens in the UI with one hand in the way we’re comfortable with our text editors.
Granted, the rollout across airlines is probably glacial
I don’t blame them. Modern UX has a huge problem with something as simple as date pickers. Preferring you scroll through 90+ items when a simple textbox would suffice.
That is a tremendously fun fact! The little background things that keep society running. May I never be cursed enough I would ever have to directly work with such a system.
It'll probably never go away, but just be layered over like civilizations. Eventually our software is probably going to get so complicated that we just build new software to interact with old software to avoid ever fully shutting it down. Like building a fresh highway on the oregon trail
I remember using some version of SABRE through CompuServe back in the day. All command line stuff over a dial-up modem, but it was novel and cool to be able to book your own flights with it. It would be very annoying to still be stuck on that interface, though.
1) Replacing any software for an airline carries huge risk. They are barely operating ok with the software they have and holding it together with duct tape. Even something you might regard as ancillary, like a baggage handling software system, or flight catering software system, if it goes down, has the potential to disrupt thousands of planes and hundreds of thousands of passengers for days.
It is so significant an issue (to try to change some software, and just one out of many systems that have to talk to each other) that if an airline ever considers doing this, they may actually stop operations for some number of days while they do it rather than risk having operations go wrong. There are some rare examples of airlines doing this to try to change their systems.
2) Related to the above, airline management hates to be embarrassed by something that might work but has the potential to go badly wrong. So they are very conservative when it comes to replacing systems that are working, even if it's painful / much less functional than what they might achieve by a change.
Combine these factors (and many others) and it means that sometimes starting a new airline is simpler than trying to fix an old one...
Sabre has a 10-year deal with GCP to do just that. It's going to take a lot more than 10 years to get the thing off System/360 running in a bunker under Omaha airport though.
The typical answer for old behemoths: it was built because it was necessary to build it, and it won't change until a change is necessary too. Wanting that change is not enough, it has to become an almost mechanical constraint, and usually the constraint gets noticed when it far outweighs the costs (and not just a little). Or is a noticeable threat to the system's existence.
GDS — there’s really only 3 centralized stores of real-time flight/hotel/booking information in the entire world (Sabre/SABRE, Amadeus, and Travelport). Almost every American airline uses Sabre (American Airlines is an interesting case in that it does not technically in a legal sense, but actually it spun off and sold Sabre in 2000, so a lot of their core systems are forks of each other)
Complexity — Fundamentally you’re looking at a logistics software, except unlike packages you’re dealing with people who aside from expected destinations have travel lengths and time-in-air calculated down to the minute. Also unlike a package, a surprise multi-day trip, unexpected multi-leg journey, one day delay is not something passengers (and crew members) will accept or be at all ok with. And if any one thing goes wrong there’s going to be cascading failures down the line— so much that it may break your company’s entire operating workflow (e.g. Southwest) entirely, and no software can overcome that kind of organizational gap.
Airlines - There’s not many commercial passenger airlines left in the US, especially that fly nationwide. Good luck trying to convince one of these giant behemoths to move to a non-battle-tested system for core operations, especially when decades-old industry software and practices around that software exist.
Entrenched - Sabre is entrenched in airlines around the world. They don’t just provide the booking services, they do the flight tracking, the ticket handling, the upgrading, the in-flight upgrades, missed connection handling, the flight scheduling algorithms, the pricing algorithms, the pilot and flight attendant time tracking, ground crew management, even the terminal software at each gate. To replace SABRE, you would physically need to rip out and then replace software around the world. And because agents don’t work from an office usually, but at the airport, you’re going to need to conduct trainings and provide support around the entire service area, which for the largest airlines is the entire world
Scale — A lot of Sabre’s revenue comes from passengers boarded. It depends on the airline, but I believe the average is that each airline pays 10cents/customer boarded with their software (though with increases in passenger volume each year, it may be less now). Because Sabre is so prevalent, and so many flights use them, they can afford such a price. A company servicing just one regional passenger airline would absolutely not be able to compete on price, at least starting out
Also— Sabre’s software itself is actually reliable! As a corporation it is slow clunky and bureaucratic, but the actual functionality it provides is stable, battle-tested, can handle any travel edge case you can think of, and fast and efficient for those who know how to use it, while also good enough at day to day operations that it doesn’t take too much time to train new agents on how to use it for routine tasks.
SABRE is ancient technology, but very reliable and at the same time extremely inflexible. Last time I saw it upclose in the early 2000's much of the core was still coded in IBM assembler, although over the decades more pieces were slowly being modernized so I have no idea where it is now. Sabre is a horrifically un- imaginative company where projects are measured in years and not much every changes.
I think though Southwest's issues are more on their side.
Yeah building a new GDS today is an exercise in insanity, it's a huge complexity nightmare and switching probably impossible. I always wondered if AI could eventually improve things, but the existing GDSs are unlikely to care much to try. It's basically a (tri)monopoly you can never break.
Great write-up. This applies to other industries such as dealerships with parts and service and various purchase plans. The software sucks for everyone except finance/accounting and that office is beside the Pres's office and therefore it won't change.
"What are the reasons preventing flight booking software modernization?"
Let's ask the opposite question:
Why do you feel that software necessarily needs to be modernized ?
I'm not sure how well SABRE works but I do know how fast and efficient keystroking through a non-GUI interface can be and I don't know why expert mode interfaces should ever be replaced by unsophisticated mousey-mouse-mouse ones.
In context, it seems like modernization simply means that the software knows whether a flight was canceled and therefore can automate the status updates for the airplane crew. Nothing to do directly with UX.
The existing flight booking software that they use (Sabre) can and does do that, but Southwest’s issue is likely their insistence on using a homegrown management system on top of that. Southwest only switched to Sabre in 2021, so it’s likely still being implemented, and their homegrown approach has likely evolved with them from their founding, so it’s not something the company itself is likely organizationally prepared for.
At least, I feel confident in that analysis, since this exact same issue happened to Southwest in 2016, before they were using Sabre. Which would point to a chronic organizational failure
airline margins are paper thin these days. they weren't back when booking engines and reservation systems were being built. (SABRE was built by and spun off from American Airlines. Galileo was built by United before they merged with Continental.) this plus the absolutely insane business logic that goes into booking engines has made the effort extremely risky.
To be fair, I think allowing flights a year in advance is probably far more complicated than just updating the underlying date format. Even if they were able to solve that problem, airlines probably can't easily operationally plan that far ahead due to so many moving parts, i.e. committing to routes and schedules, planning for staffing that far ahead of time, ever changing government restrictions, fuel price fluctuations, inflation, geopolitical realities, staffing, etc. I mean, imagine if they did, and something like COVID comes along again, it would cause far, far more disruption if they had booked out the next few years in advance (we had no idea how long COVID restrictions would last while we were in the heat of it, it's only clearer now in retrospect).
Also speaking as a software engineer myself, it's almost never just a software fix that will magically solve everybody else's problems, that always ends up being just wishful thinking
While this is humorous in that there are limiting assumptions like this baked into the system, I also have to wonder, who needs or even wants to book a flight a year in advance? I dread planning out a flight 4 months in advance and dealing with the almost inevitable cascade of conflicts this introduces of juggling and rescheduling things to make things align correctly. One year makes me cringe.
There are events that I can see purchasing flights well in advance for. I used to a go to a conference that was held every other year at the same time, it would have been easy to buy tickets more than a year in advance for that without much concern. Eclipses, certain sporting events, or reservations for activities with a wait list of more than a year could qualify as well.
Despite that I am like you and rarely have tickets far in advance of a trip.
- Annual conferences or conferences that occur every-other year
- Planning family reunions because you need that kind of cat-herding lead time when you have 9 uncles/aunts on just ONE side of the family
- Periods where I have some spare cash I'd like to lock in a getaway with before I spend it or something unexpected like the invasion of the Ukraine drives up fuel costs and overall prices... or a global pandemic hits - would be sweet if I could have rebooked some of my trips for for 1-2 years out when the pandemic hit
- Travel for future medical stuff; at one point for 2-3 years I was taking my mom to the Cleveland Clinic every 4 months for periodic checks and it would have been super nice to be able to just book that stuff way in advance and have it all taken care of
Etc
etc
etc
I'd bet quite a few people would appreciate that ability
My family plans the yearly family get-together at the yearly family get-together. A year in advance. Except sometimes due to scheduling deconfliction, it's actually 10 or 11 months in advance.. or 13 or 14 months in advance. The exact date floats and sometimes we are planning trips more than a year in advance.
I could see it for major holidays. I spent too much money to fly home this year because I am bad at scheduling. I would consider booking next year's flight during this year's trip just so I know it's knocked out and I don't have to worry about it.
,, Most of the travel industry runs on old software that would horrify a lot of people here''
If you can see how it works, it horrifies me even more as a traveller, as from outside it just doesn't work a lot of the time.
Also if you just look at the video, we all know how bad these systems are, but are not able to do anything (starting anything new in the airline industry has too much cost).
Likely cannot book that far not because of the underlying date format, but because of jet-a fuel prices which fluctuate. Airlines typically hedge their near term purchases with longer-term futures
Most of the travel industry runs on old software that would horrify a lot of people here, especially those who've never worked for a large, 30+ year old company. When I used to interview a lot of people I made it a point to mention some of the more "interesting" aspects so they'd know what they were getting into.
One example: ever tried to book a flight a year in advance? On a lot (almost all?) of systems you can't, because the underlying date format is "DEC27".
Edit to address a couple comments: logistics are hard and there are plenty of reasons why airlines wouldn't want to support booking that far out. However, the reason you can book a flight 330 days from now but not 360 days from now is almost certainly due to the date format. (I believe the windows used are less than 365 days because it's helpful to be able to have dates in the recent past. I remember seeing documentation for 360, but AA and United seem to be in the 330-340 range on their websites).