This Tech Bubble Is Different

The critique that web companies don't leave a technical legacy is not really valid. Most businesses don't leave a legacy.

Just because a lot of time is spent working on "non-interesting" problems, technologically speaking, like Facebook and Zynga, doesn't mean that there aren't plenty of startups who are working in more technologically-impressive fields (). The entire article is completely anecdotal.

To make things worse, it explains how Groupon's only tech innovation is "cute email". But I don't understand why people call Groupon a (technology) startup. It's just a company. A fast growing one, sure, and that relies on some technology (like every other company in the world). But it's not a tech company at all.

I'm not saying that Facebook isn't impressive, technologically. I think Facebook has a lot of things to offer, e.g. scaling, building the world's largest real-time chat system, impressive amounts of work into user interfaces, and, as the article mentions, impressive amounts of work into analyzing user behavior.

Hadoop was developed at Yahoo, an advertising supported company. Some of the most interesting contributions (HDFS sync, RaidNode, AvatarNode) have come from Facebook, an ad supported company. Hadoop's architecture itself is inspired by the MapReduce and GFS papers from Google, another ad supported company. This shows that the premise is false: there is indeed technology transfer from the "hot businesses" to other companies and academia. Hadoop is the prominent example, but not the only one: many contributions to the Linux kernel come from Google, both Google and Facebook have also contributed extensively to MySQL.

The ad ranking and recommendation algorithms themselves are also by no means trivial, much of the work that goes on in that field (machine learning, information retrieval, natural language processing, large scale graph processing) is applicable elsewhere. The "math wiz" college graduates working on these fields, are going to be able to apply the skills they build elsewhere. Not to mention, the stock options, even if we assume modest outcomes, could pay for many a Ph.D. for engineers who wouldn't otherwise afford to take the time off from industry.

There are also products like GroupOn that don't seem to be technology driven at all, but that can change rather radically (e.g., Amazon's move from a technology consumer, to a producer of their own technology, to a company selling technology to others).

In other words, one doesn't have to sell technology to businesses in order to build technology that benefits others. There's nothing wrong with selling technology to businesses (and if I were to start my own company, I'd start one that does that, because I understand e.g., distributed databases and market for them better than I understand, e.g., machine learning and consumer marketing), but there's nothing wrong with providing an ad supported service to consumers, as long as interesting and universally applicable technology gets built in the process.