3 ms·

The annoying part about booking plane tickets these days is that most flight booking websites have very similar and limited functionality for querying and filte

The annoying part about booking plane tickets these days is that most flight booking websites have very similar and limited functionality for querying and filtering. Sure, there's Matrix ITA and Google Flights is an upgrade too. But imagine what people could do if you gave them full fledged access to all flight data. The efficiencies!

squeaky-clean 5y ago
A lot of flight prices are calculated upon a search query and only guaranteed for that query (and for a set number of minutes). There generally isn't just some sort of table you can look into and get prices unfortunately. The company I work at provides these sorts of search interfaces to airlines, but all the pricing data we show has to be estimates based on confirmed prices in the past. If you're directly on an airline website (not Google Flights, Kayak, etc) and see a calendar or something that lets you view 60 days of prices or more at once, there's a good chance that was developed by one of our teams. Unfortunately we can't totally guarantee you those prices, and we'll do things like not render those components if our estimate confidence isn't high enough. Getting a "real" price for a flight is not a particularly fast API call, and also costs the airlines a small amount of money from the provider (almost no airline directly generates their own prices, generally another company does). Fractions of a penny per call, but that really adds up if you just give free bulk direct access to that to customers.
- OJFord 5y ago
  Is that not all totally bonkers?
  - splonk 5y ago
    People who don't work in travel would be horrified at both acceptable latencies in that space and just how old some of the formats involved are. One example I used to tell people in interviews: ever tried to book a flight more than a year out? You probably can't, because the date format in a PNR is "APR29". That's the underlying data for basically all flights.
splonk 5y ago
There's a paper somewhere (I think from one of the ITA founders) about how flight search is actually NP-complete. I haven't looked carefully but I think these slides cover a lot of it. http://www.ai.mit.edu/courses/6.034f/psets/ps1/airtravel.pdf http://www.ai.mit.edu/courses/6.034f/psets/ps1/airtravel.pdf "One interesting result not written up here is that even completely fixing the flights and fares of a ticket, so that the only remaining question is how to partition the fares into priceable units, is NP-complete. This is interesting because only flight and fare information makes its way onto printed tickets, not the grouping of fares into PUs. Therefore the problem of just validating a printed ticket is worst-case NP-complete, though it is rarely difficult in practice."

squeaky-clean 5y ago

A lot of flight prices are calculated upon a search query and only guaranteed for that query (and for a set number of minutes). There generally isn't just some sort of table you can look into and get prices unfortunately. The company I work at provides these sorts of search interfaces to airlines, but all the pricing data we show has to be estimates based on confirmed prices in the past. If you're directly on an airline website (not Google Flights, Kayak, etc) and see a calendar or something that lets you view 60 days of prices or more at once, there's a good chance that was developed by one of our teams. Unfortunately we can't totally guarantee you those prices, and we'll do things like not render those components if our estimate confidence isn't high enough. Getting a "real" price for a flight is not a particularly fast API call, and also costs the airlines a small amount of money from the provider (almost no airline directly generates their own prices, generally another company does). Fractions of a penny per call, but that really adds up if you just give free bulk direct access to that to customers.

OJFord 5y ago
Is that not all totally bonkers?
- splonk 5y ago
  People who don't work in travel would be horrified at both acceptable latencies in that space and just how old some of the formats involved are. One example I used to tell people in interviews: ever tried to book a flight more than a year out? You probably can't, because the date format in a PNR is "APR29". That's the underlying data for basically all flights.

splonk 5y ago

There's a paper somewhere (I think from one of the ITA founders) about how flight search is actually NP-complete. I haven't looked carefully but I think these slides cover a lot of it. http://www.ai.mit.edu/courses/6.034f/psets/ps1/airtravel.pdf http://www.ai.mit.edu/courses/6.034f/psets/ps1/airtravel.pdf "One interesting result not written up here is that even completely fixing the flights and fares of a ticket, so that the only remaining question is how to partition the fares into priceable units, is NP-complete. This is interesting because only flight and fare information makes its way onto printed tickets, not the grouping of fares into PUs. Therefore the problem of just validating a printed ticket is worst-case NP-complete, though it is rarely difficult in practice."