In recent weeks, the media is having a field day with Marissa Mayer’s most recent acquisition of the 6-person Summly team out of the UK. With a teenage cofounder, it looked initially like Yahoo had acquired a great new technology, or perhaps a stellar team. But as word comes out that only 3 of the team members will be joining Yahoo, doubts about the $30 Million acquisition have begun to arise. While it seems now that it must be a technology – or, as NPR called it, a math acquisition – some say that the algorithms that power summaries like the ones provided by Summly are just not right, and that it requires a human touch to get right.
One such critic is tl;dr, a Paris-based company which recently got accepted into Seedcamp, which provides a platform to allow internet users to provide “tl;dr versions” of articles (that is: “too long; didn’t read”). The user-generated summaries can be viewed via a Chrome Web App which allows you to quickly see if any given web page contains a summary, as well as to click and see the summary itself. Through its curated summaries, tl;dr CEO Stanislas Marion hopes to summarize the internet, one page at a time.
I first met Stan last year at Index Ventures’ rooftop party on the Criteo Paris office, and initially I was skeptical; however, I’ve been using the application for quite some time – not so much as a summarizer but as a consumer of summaries – and I’m starting to see the value. While the company is going to have a long haul to hit critical mass and begin seeing the summaries flowing in on a daily basis, Stan says the company has been growing 10% per week since the beginning of the year, and 25% since the Summly acquisition.
Stan argues that, while taking ‘smart snippets’ of text might pull the most relevant keywords out of a given article, it’ll never provide as much value as someone who took the time to read the article, and can paraphrase something to a much-more understandable summary. I for one would be happy to see some sort of tl;dr WordPress plugin that would let blogs automatically generate their tl;dr versions – after all, the writers themselves might be best fitted to give a tl;dr version of their own rants.
It seems that Stan isn’t the only one thinking like this – Om Malik’s recent blog post on Data Darwinism discusses the same topic and even gives mention to tl;dr themselves.