Awesome. I can never understand what antiquated database and UI the other flight reservation websites are using. Simple database queries take 10+ seconds, and then you have to click N more times to get the taxes included.
Why can't I query for
SELECT * FROM tickets WHERE sqrt((destination_latitude-desired_latitude)^2 - (destination_longitude-desired_longitude)^2) < 5000 AND abs(unix_timestamp(departure)-unix_timestamp('somedate'))<10*86400 ORDER BY (price+taxes+fees);
with proper geoindexing and sharding this query should take no more than a fraction of a second, yet airline websites want me to use some stupidly-designed UI that can't implement above query, takes 10+ seconds to return results, and doesn't include the taxes and fees. Thus I need to often manually enter every destination within 5000km and every date combination and manually waste half a day to find the cheapest ticket to "anywhere interesting".
Furthermore often times changing the language and country of the website changes ticket prices; a well-designed, globally-consistent database should not allow this to happen.
I hope that Google will disrupt and reengineer this ancient system.
Most companies don't want to take on this complexity, so they outsource it. The companies that provide this service, like ITA, tend to charge more money for more expensive queries, like multi-day searches, so most online search companies will only let you do the cheaper kinds of searches.
I wonder if it would be viable to start an airline whose business model is "simple pricing". Just forget ITA altogether, sell tickets directly, and have the pricing model extremely simple. Like, total price = C/(time to departure) for some constant C, and use machine learning, or just a simple PID feedback loop on revenue to set C in order to be profitable.
As the presentation above suggests, this would have a problem with, well, staying profitable. Tourist air travel is so heavily underpriced that you can barely tank up the jet with what you collected from people. That's one of (or maybe even the) reason for such variety in ticket prices.
Airlines spend a absolute fortune on complex pricing for a reason... they wouldn't survive (in the absence of protectionism or a natural monopoly on niche routes) without it. The optimum "one size fits all" price doesn't exist.
I don't think this would work unfortunately. Players in air travel face a) intense competition and b) tiny profit margins. This forces some 'creative' pricing.
For the same reason I can't just renew a domain at Network Solutions without going through eight different pages trying to upsell me things (and often defaulting to checking on those services).
Expedia, Orbitz, etc make money via nag screens and upsells. They have no incentive to make the experience positive or easy. Google seems to.
I've switched from 1&1 to gandi.net - and while they are a little more expensive it's been pretty smooth. Also they support TOTP which is a huge awesome in my book (I used TOTP with facebook, Microsoft account, google, source code committing etc).
I'm convinced that for every DNS related change at 1&1 - there is a guy that manually makes the change.
I switched to Namecheap; they sell stuff but are not annoying about it.
Google Domains also looks promising but I'm waiting for them to not be in beta lest they decide to scrap the product. Google would win me over if they offered free SSL certificates. In fact I think they should go ahead and disrupt the SSL industry by doing that too. Nobody should have to pay for a silly cryptographic signature. That would disrupt NS and all the others who try to upsell stuff.
> […] if they offered free SSL certificates. In fact I think they should go ahead and disrupt the SSL industry by doing that too. Nobody should have to pay for a silly cryptographic signature.
> ...free SSL certificates. In fact I think they should go ahead and disrupt the SSL industry
StartSSL already offers free SSL certificates - and they haven't disrupted the industry. Even if Google started offering free SSL certificates - it's only a band-aid to the problem not a solution.
Have you tried using StartSSL??
(a) They aren't free if you are doing anything 'commercial'
(b) I decided to try and give them my money. I tried very hard. I gave up.
After uploading your documents 30 times, them claiming they didn't get them, you would too.
Oh, and uploading them each time was pretty painful.
Yeah, this is terrible engineering design IMHO. The providers should just provide some kind of firehose of updates to the booking websites which can then cache the data appropriately.
Also they should fix their cache implementations. I've had countless times on the AA website where it says a ticket is available; then when I try to book it it says it's not available; then when I go back to the home page and search again the same fare shows up again. Seriously, where do their programmers come from ...
There are basically an unlimited number of world wide routes - a mind blowingly dizzying number of permutations to get from A to B.
Before you try to say "well travellers won't want routes with feature X" to cut down on the volume of data travellers want all sorts of contradictory things. Some a vastly price sensitive, some hate a particular airline, some love that same airline, some are time sensitive, some change those preferences depending on whether they are flying domestic or international, with or without kids.
> a mind blowingly dizzying number of permutations to get from A to B.
Yes, but I don't think that's what most people desire to enumerate (or what the parent of your post had in mind). A simple enumeration of the flights available and their one-way prices would be a start. My understanding (from a link in another post) is that this would be ~(1 flight) * (average number of changes to a flight's data between birth and flight) updates per second, which I think is within reason.
My understanding is that airlines frustrate this by not pricing just individual flights, but giving odd discounts on round-trips and other odd combinations (such as A→B C→A) of flights, meaning you need more than just one-way prices. (There's also the fact that some airlines have like 5 or 6 different prices for a seat all on the same flight, depending on where in the plane the seat is.)
There's all kinds of reasons that this isn't sufficient, return based pricing, multiple seat pricing bands for a single flight (for some airlines this is dynamically calculated on a moment-by-moment basis), airline and airports rules covering transfers for whether a given pair of flights connect or not, etc... the list is basically unlimited.
Yeah, well, if you can get providers to agree to do anything at all, let alone have the capability to not be in the engineering dark ages, you'll be rich. Until then, companies will continue to fill the gap with imperfect solutions.
So we need an API to search all the languages and countries, pay in the appropriate currency, abstract this "feature" away along with the language barrier, and give me the lowest price. I hope Google will do that. :)
It would likely break all sorts of TOS agreements at various levels, and wouldn't last a month.
There are industries where profits are huge and disruption is badly needed. Airlines are not one of them, afaik; ever since low-cost became a thing, they've been under constant economic pressure and usually operate on razor-thin margins, at least in Europe. Mid-size companies have all but disappeared, even giants like British Airways and KLM have been permanently de-structured.
Flying is an expensive and risky business, piling up the pressure is likely to generate even more safety issues than we already have today. If you're all about price, just fly with RyanAir and be sure that nobody can ask you for less money than they do.
Oh the big consolidated IATA airlines are raking it in, don't worry about them
"On a per passenger basis, [IATA] airlines will make a net profit of $7.08 in 2015. That is up on the $6.02 earned in 2014 and more than double the $3.38 earnings per passenger achieved in 2013."
Decentralization to the rescue? Create a decentralized system by which such scripts can be circulated and updated without anyone owning them.
In general I live by (0) anything that can be scripted and automated should be automated and (1) humans and machines are no different and the boundary between human and machine will blur in the future, so we might as well stop differentiating the two now, assume the customer is infinitely smart, and define all policies and business models using only initial state and final state of the customer rather than the process the customer went through to reach that state.
For example, my library books can be renewed online (or else one pays a fine). This can be automated. Thus, I write a script to abstract this problem out of my life. The library need only care when the loan out and when they get back, everything else about the system including whether it's a collection of silicon components or a collection of sodium-potassium-pump components that causes the state transitions at book renewals is none of their business. If I invented something that used germanium to cause a renewal (and biologists can't figure out whether to define it as living or nonliving, intelligent or nonintelligent), the library shouldn't care.
Similarly, it's none of the airline's business whether it's sodium-potassium, silicon, or germanium that transfers entropy about tickets; the initial and final states are the same. if I can gain information about airline tickets by manually querying 1000 times from different parameters, that process should be scripted and automated.
I'm a machine just like my computer, just a much slower one.
Why can't I query for
with proper geoindexing and sharding this query should take no more than a fraction of a second, yet airline websites want me to use some stupidly-designed UI that can't implement above query, takes 10+ seconds to return results, and doesn't include the taxes and fees. Thus I need to often manually enter every destination within 5000km and every date combination and manually waste half a day to find the cheapest ticket to "anywhere interesting".Furthermore often times changing the language and country of the website changes ticket prices; a well-designed, globally-consistent database should not allow this to happen.
I hope that Google will disrupt and reengineer this ancient system.