Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Does anyone know what makes ITA so special? Why don't any of the big airlines have APIs that Hipmunk, Kayak, Orbitz, etc. can use directly? (Do any of them?) Seems odd that everyone should have to rely on this one company for the data, when it should be originating from the airlines.


ITA's strength isn't data, it's algorithms. Finding the best flight is not just NP-complete, it's uncomputable. In practice, even a lot of sub problems are NP-complete.

http://www.demarcken.org/carl/papers/ITA-software-travel-com...

People think the problem is a graph search problem of finding the shortest path. But that's actually the easiest part of the problem. It turns out that prices aren't attached to individual flights, but rather are rules of the form "If the passenger goes from A to B on a Tuesday, the price is X, and it doesn't really matter what flights they take, as long as the total distance is less than 3 times the line of sight distance."

Then there are all sorts of rules about what fares can be combined with other fares, under which conditions, ...

I worked at ITA software on QPX, and their code is by far the most algorithmically complex code I've ever seen. As I mentioned, there are many, many sub problems, and each one is a tricky algorithm design problem.


Orbitz and Kayak both use them, check the wikipedia page on them for more info in that area[1]. ITA's real claim to fame, at least in communities like HN, is that they're one of the largest Common Lisp shops still around. Also my experience interacting with their employees has been that they're an exceedingly smart and talent bunch of people. Additionally, and especially if you lived in the Boston/Cambridge area, they're pretty well known for there rather challenging interview puzzles which used to be plastered all over the T [2]

[1] http://en.wikipedia.org/wiki/ITA_Software

[2] http://www.itasoftware.com/careers/puzzle_archive.html


I answered part of this question earlier at http://news.ycombinator.com/item?id=1473748

The thing to understand is that the airlines publish data that ITA uses. The problem is that it takes an incredible amount of engineering effort and algorithm smarts to go from 'published airline data' to 'this is the cheapest flight from BOS to LAX on 4/20'. ITA does have competitors; historically those competitors offered much lower quality products, although that might have changed.

One of the ITA co-founders wrote a presentation on some of these topics at http://www.demarcken.org/carl/papers/ITA-software-travel-com...


ITA is special not because of the data but because of what it does with the data. The others haven't been able to reproduce that. Think of it as the Google of airline data.


To give an example, check out http://matrix.itasoftware.com/

I've never found anything as useful or flexible for searching airline data.


> Think of it as the Google of airline data.

And now they will be the Google of the Google of airline data.


The data is available (not cheaply) from an company called ATPCO, which is owned by the airlines. The airlines publish fares and rules through ATPCO, and ITA, SABRE, and many others fetch the data and run the computations.

The computations are the hard part.


So if Hipmunk, Kayak, or whoever wanted to remove their dependency from ITA (by doing more computational stuff themselves), what's the best place for them to get their data?


I'm pretty sure ATPCO is the only way to get the raw data; people who subscribe to it are not allowed to redistribute it.

As for getting processed data, that's called metasearch: scraping other sites' search engines. Kayak started out as pure metasearch, but moved to using ITA. Not sure whether they still do any scraping.


Herez my post on why google needed to buy ITA. http://bit.ly/h07ked


here's the actual link that the bit.ly link directs to:

http://www.thoughtsonconsumerweb.com/why-google-needs-to-bui...


While your thesis may be correct, some of your assumptions are a bit off:

1) Bing has stopped investing in travel tech, they now use Kayak for search. So do we still believe in the value prop?

2) Farecast (fare predictor) turns out to be not very valuable. So much so, that Bing couldn't even sustain development of their own search tool. Can Google do a better job of it? Maybe. Can they do a good enough job of if that it's ROI positive? I'm skeptical.

I agree that Google believes many of your assertions. I don't agree that this makes it true. Do people really need more sophisticated travel search functionality? Sure, you can imagine lots of cool stuff -- bells and whistles. But ask Kayak if their "Explore" feature (perhaps a lite version of what you can imagine Google doing with ITA) drives their business? I'd argue it's neat, good for branding and loyalty, but not at the end of the day, a major source of value.

That being said -- the key thesis of: Google trying to hold onto relevance in travel search -- is true. However, they are still pretty much a one-horse show (search revenue), and travel is one of (if not the) biggest vertical today.


Microsoft butchered Farecast, it never really had a chance there.


Citations?

I didn't know Bing was using Kayak now, and Farecast has been quite useful for me (fairly accurate) in the past.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: