Google buys ITA Software (Part 1: The back story)

At the start of this month Google announced its purchase of ITA Software — if the antitrust division of the USA Department of Justice, and/or the Department of Transportation, decide the deal won’t restrain trade by giving Google any sort of monopoly — for US$700 million in cash.

Even for Google, this is a major purchase, and there’s been lots of attempted analysis of how it will affect the financial fortunes and who the winners and losers will be among competitors of Google and/or ITA Software — a task complicated by most investors’ and even financial analysts’ ignorance and misunderstandings of the travel technology infrastructure, which has led to considerable debate and confusion as to what business ITA Software is in, and with whom they compete.

The best of the industry analysis has been in online trade journals TheBeat.Travel (paid subscribers only) and Tnooz (public; check more recent articles for continuing coverage). As Tnooz was perhaps the first to point out, Google itself made clear when it announced the deal that it saw ITA Software as part of a layer of technology intermediaries consisting mainly of the Computerized Reservation Systems or Global Distribution Systems (CRS’s/GDS’s): Sabre, Amadeus, and Travelport’s Galileo and Worldspan divisions. But because CRS/GDS companies are generally invisible in their intermediary role (and currently all [except one] owned by groups of private equity investors, so they need not report publicly on their finances or operations), few analysts outside the travel tech industry know how to interpret the implications of Google’s decision to invest $700 million in this sector.

[Correction: An anonymous commenter on Slashdot points out that Amadeus, one of the four major CRS’s, began trading publicly on Spanish stock exchanges on 29 April 2010. In my linked article about CRS’s, I had mentioned that such a public offering was pending, but hadn’t noted the IPO. The largest fraction of Amadeus stock continues to be held by private equity firms, however.]

Frankly, I’m not at all sure Google itself understands what ITA Software does (and doesn’t) do, and what they are getting for their money. But that’s getting ahead of myself, and largely irrelevant unless you are considering investing in Google or the travel industry.

The missing element in the discussion to date, and the question I’ve been getting from readers, is, “What will this deal mean for travellers?”

The short answer is that it is likely to be a bad thing for travellers, not because it will give any company a monopoly over anything it doesn’t already dominate but because it is likely to exacerbate the trend toward personalized and less transparent pricing of airline tickets (and other travel services) and the de facto disappearance of key consumer protection principles embodied in the definition of a common carrier and the requirement for a published tariff applicable equally to all would-be customers complying with the same rules.

The long answer starts with the largely unreported back story of ITA Software’s origins, as a predicate to understanding how its acquisition by Google might affect travellers. I’ll tell that story in the extended article below, and explore what ITA Software actually does and the implications for travellers in Parts 2 and 3 tomorrow and Wednesday.

ITA Software began with the friendship between Jeremy Wertheimer, then a doctoral student in computer science at MIT, and Richard Aiken, a (currently non-practicing) dentist and serial entrepreneur who had developed a side business as a travel consultant, using his largely self-taught air tariff expertise and connections to some of the key figures in the then-emerging USA consolidator and bucket shop industry to find cheaper international air tickets for his circle of Cambridge acquaintances.

Knowing some of Aiken’s customers, Wertheimer was impressed with Aiken’s results, but equally impressed with his own ability to solve hard problems through computer algorithms. ITA Software began with the idea that Wertheimer would replicate Aiken’s airfare expertise and heuristics in “expert systems” software. To that end, Aiken began giving Wertheimer an intensive series of “brain dumps” of his knowledge of how airfares and airline ticket pricing work, and eventually they formalized their partnership with the incorporation of ITA Software.

Wertheimer and Aiken are both, as Wertheimer likes to say, extremely “clever”, but it’s no surprise to anyone who’s met them both that there wasn’t room for both their egos in one company. Wertheimer says that the whole idea of a heuristic approach was quickly revealed to be less effective than brute force implemented through sufficiently clever algorithms and efficient programming, and Aiken’s air tariff and industry expertise didn’t prove to be necessary. Aiken says Wertheimer picked his brains and then forced him out, relying ever after — even for Wertheimer’s brute force solutions — on the essential foundation of the crash course Aiken had given him in how the system of airfares works and in the air travel IT ecosystem.

To the extent that there was a substantive difference of opinion and not just a personality clash or power struggle, the strategic dispute between Aiken and Wertheimer mirrored the historic split between “brute force” chess-playing computer programs and heuristic programs that try to mimic the strategic thinking of human chess experts. Most of the latter category have been developed in partnerships between chess grandmasters and expert systems programmers, and have been successful in proportion to the expertise both of the human experts and the programmers.

(I was part of such an expert systems software development partnership as an air tariff and industry expert working with programmers hired from outside the travel industry at Airtreks.com, with which I am still affiliated. By way of full disclosure, some of Airtreks’ patent applications were denied because — from my point of view — the patent examiners didn’t understand enough about industry terminology and the differences between published “fares” and consolidator “prices” to distinguish Airtreks’ innovations from software patented slightly earlier by ITA Software.)

One way or another, by 2000, Wertheimer was writing Aiken out of the corporate history, telling Business Week that, “for years, the company was nothing but Wertheimer and an MIT chum”. (I’m not sure if that “chum” was Dave Baggett, who was similarly disparaging of travel agents’ prior art, or Carl de Marcken, who was awarded most of ITA Software’s patents and in whose name other ITA Software patent applications remain pending.)

Wertheimer and Aiken wound up in court arguing over the meaning of their oral agreements about the terms of their partnership.

Aiken continued to follow ITA Software’s fortunes from afar while moving on to other entrepreneurial ventures including an indoor climbing gym on International Drive in Orlando, Florida, Adrenaline Heights, and an abortive attempt to set up a similar venture in Manhattan. When I contacted him last week, after the announcement of the Google-ITA Software deal, Aiken told me that eventually he had settled with Wertheimer, and retained some stock in ITA Software. But my assumption is that his share in the windfall from the Google purchase is small, and his role and influence on the current direction of the company nil.

After Aiken’s departure, ITA Software shifted its focus from international to domestic USA airfares. Aiken says that’s because international pricing was (and is) more complex, was less amenable to a brute force approach, and would have required more of the expertise he had been providing. Wertheimer admits that international pricing and routes are “interesting” and “hard” (perhaps the highest compliment in his vocabulary) problems, but says that it was purely a business decision, driven by the needs of potential customers, to tackle the domestic air travel puzzle first.

So far as I can tell, and so far as Wertheimer has been willing to disclose in response to my periodic questions, it wasn’t until several years later — well after the launch of ITA Software’s domestic airfare “search” product and its deployment by Orbitz.com, and when ITA Software was finally (by customer demand, Wertheimer said) trying to expand its functionality to encompass international flights, that they hired their first employee with air tariff knowledge. Even then, their role seems to have been limited to explaining how fares and rules work, rather than being a source of expertise, insight, or heuristics for the programmers to emulate. If all that has come before is primitive, what point would be served by trying to learn about its primativism or hiring any of the Neanderthals who’ve built and operated it?

Today, the most striking thing about ITA Software may be the continuing absence from its staff of any more than the absolute minimum of people with any prior experience or expertise in airline pricing or airline IT.

What has ITA Software actually done? And aside from the corporate soap opera (why aren’t more sitcoms set in the hothouse of a high-tech start-up, or am I missing something?), why should travellers care about any of this? I’ll talk about what ITA Software does tomorrow in Part 2, and the implications of the Google-ITA Software deal for travellers Wednesday in Part 3.

[Update: There’s more background on how airline ticket pricing works in my books, and more comments on Slashdot.]