HipMunk just announced they will be shutting down next week. Back in 2010 when they first launched, they were by far the best site for searching for flights. They also had the chipmunk mascot – who people have told us strongly resembles RentHop’s Wally the Wallaby.
Turns out they are brothers! Or long lost cousins, or something. Alexis Ohanian created both, back when he was the Y Combinator Ambassador to the East. Obviously NYC is home to many startups now, including YC Alums. But back in 2011, there was still huge bias towards staying in the Bay Area after graduating from YC. Office hours and all the investors and other founders were still near Mountain View (or SF), so moving back to New York meant giving up many of the alumni network benefits.
The Early NYC Years For YC Alums
RentHop was among the first to lead the charge to New York, along with a handful of the old guard: Octopart, OMGPop (Draw Something), The Muse, Amicus, and TutorSpree. Aaron Harris’s TutorSpree at the time hosted Alexis Ohanian and Kat Manalac in their sweet SoHo office spaces, and there we would have our office hours which led to Wally’s creation!
HopScore and AgonyScore
Probably the best piece of advice we got from Alexis was to organize all of our data and search results in a way that was very useful for consumers. He pointed to the innovative AgonyScore, that HipMunk used with great success. Back in 2010, no search engine sorted flights in a sensible way (and no apartment site sorted listings sensibly). They realized travel is painful, but they could find the least agonizing flight possible. Good flights take off and land at desirable times and minimize craziness with transfer. Almost everyone would pay just a little bit more if it meant far more convenience. HipMunk would also suppress flight options that were obviously bad (same price but longer layover), and they were the first to clearly display WiFi capabilities on the flight (back when in-flight WiFi was usable — it seems to be getting worse).
Around that time, we created the first version of the HopScore, using many of the same principles. Not EVERYONE values the same things. Maybe some people like red-eye flights. Maybe some people appreciate the pre-war charm over the modern highrises. But a basic algorithm and score should apply to the majority of searchers. It is also possible to personalize the score based on past behavior or stated preferences. But the big first step is simply gathering and aggregating all of the possible features.
HopScore Today
The HopScore continues to evolve in many ways, year after year. Modern machine learning techniques downplay the importance of engineering “features” in your data. Instead, the deep learning proponents encourage fitting enormous amounts of parameters to large data sets. Regardless, we credit HipMunk for a big part of our initial inspiration designing the interface and back-end of the HopScore. It will be sad to see them go!
