Hacker News new | ask | show | jobs
by endofcapital 2655 days ago
"But Google Fiber got something out of its time here. It learned that nanotrenching—the cost-saving process of burying fiber optic cables just two inches underground—was a bust. “We currently do not have plans that call for 2 inch trenches, our primary specifications are focused on going deeper,” a Google Fiber spokesperson said in an email."

It's so weird to just use cities as A/B tests and just disregard all the people and plans built around a failed case at the drop of a hat. Are they going to start A/B testing countries against eachother next?

4 comments

Trying new products in small markets as an experiment has been a thing forever, and is something literally every company does regularly. There is nothing weird about it. As for the fallout, IMO the city is to blame for it more than Google, since the company is obligated to do exactly what the contract entails.
Yeah, but when a soda company, say, tries out one new flavor in my town, and different new flavor in your town, and mine flops and yours succeeds, they don't just stop selling soda in my town. They start selling the flavor from your test in my town, too.

It would have risked much less bad publicity if they had done the 2" experiments in a city they were willing to use 6" trenching in. Do part of the deployment with 2" and part with 6", and if the 2" part doesn't work redo that part with 6".

Soda companies also don't leave deteriorating trenches behind on the roads in town when their tests fail.
I think what's different is typically cable/phone companies have been reluctant to give up on their investment so easily. This is probably somehow due to the past and present monopolies they hold. Google, like the honey badger, don't care.
This already happens, Facebook sometimes tests features in specific countries e.g. New Zealand before wider rollout.

https://i.stuff.co.nz/technology/social-networking/103511194...

Isn't that what cities, states, and countries are in an abstract sense? They are randomized trials of different systems of law and government, and so on.

It's much messier when it's accidental, but ... wouldn't it be smart for say, new state propositions in California to be tested in a trial to verify they produce the consequence voters asked for?

Google does this all of the time with all of their products. The purposely degraded the performance of some users on their products (search IIRC) to see how they would react.
YouTube too, can't find it now but they did a study where they degraded users to 480p to see how many would manually switch back to HD (spoiler: very few did)

I believe YouTube also experimentally surfaces new recommendations slowly to see how many clicks they get and if it should be surfaced more or not