It fails basic civil engineering on so many levels.
This is classic HN armchair engineering. I'm sure Google didn't put any thought into their physical rollout.From the article:
> “If anyone understood road technology or tire technology, [they] would know it’s going to be a problem,” said Jim Hayes, president of the Fiber Optics Association and a decades-long veteran of the industry.
> “They bid out the work to replace the sealant with asphalt,” Simrall said. But before the work started, Google Fiber saw customers lose service when the process of repaving roads damaged fiber lines. “So they shifted their attention to addressing that,” she said.
Apparently, they did not in fact put any thought into it, besides "how can we cut costs".
> Is cold weather bad for fiber?
Both the Frost Line wikipedia page I included and the Frost Heaving page that has been also posted paint a picture, but if you really want a lay description, here's a rough one:
The frost line is about how far down groundwater seeps into the soil (and even most types of asphalt) and generally remains there in some proportion throughout the year. One way to think of it is that the ground can be considered some form of "moist" at and above that line at just about all times of year. That hugely comes into play in cold weather which is why it is called the "frost line": water expands as it freezes, it takes up more space in the ground and in the asphalt. This expansions causes "frost heaving", the expansion of cold water causes things in the ground to sometimes not stay in the ground, chunks of asphalt don't stay concreted to the rest of the road (potholes), etc.
So the frost line is something of the line of "if you want a thing to stay buried, you must bury at least this below the frost line". That includes buildings as the frost line even determines some of the safety requirements for the foundations of buildings as they too can get frost heaved out of the ground in worst cases.
There's never a case where "nano-trenching" 2" into roads makes sense when 2" (or more) potholes are common naturally in cold weather. That's why civil engineers have used terms like "frost line" for centuries at this point. Just one pothole and the wire is at the very least exposed to the elements. At that point it is running through the middle of a pothole in the middle of a road so you should be able to guess what the most likely next outcome is, even in the best case that the wire survived the frost heaving.
(This is also before you add in the fact that burying under 2" of asphalt wasn't even their first attempt, but their second. Their first attempt was burying under 2" of epoxy or as more commonly named: glue. Their terrible epoxy melted in warm weather, creating warm weather potholes in the streets. Their terrible epoxy mixed worse with water and turned city streets into nothing but potholes in the cold weather.)
On its own, it's really hard to give Google any credit for forethought in their rollout because potholes in cold weather is such common sense, no shit, Sherlock it shouldn't have ever needed to be said up front.
That said, I'm sure it was said up front a lot that the city was "untrenchable". Most of the city, especially the urban core and several major neighborhoods, isn't a good candidate for any sort of trenching. Not just because the city has a low frost line (which it does), but also because the city is in the middle of what is referred to as "karst country". The city is on top of giant limestone deposits which are full of caves both man-made and natural. Dig too deep (which is not far below the frost line of the city) in many neighborhoods and you risk sinkholes and cave-ins and worse.
A civil engineer I'm not allowed to name because they were under NDA at the time and Google would sue them said to me once, and I paraphrase because there was a lot more cursing and foul language over beers "What part of untrenchable did those idiots not understand?"
Isn't it just that traditional trenching would have been too expensive for Google to justify?
Google apparently didn't even want to spend the expense running it on utility poles, which frankly, I would naively think would be the cheapest option.