Hacker News Reader

I like the idea from the point of view of 'making things better', but from the philosophical perspective of publicly funded entities should not give benefit and preferential treatment to private and proprietary enterprises and systems.

I think if the USPO was capturing this data, then it should be available to everyone.

>publicly funded entities should not give benefit and preferential treatment to private and proprietary enterprises and systems.

It wouldn't be a "gift" from the USPS. It would be a mutually beneficial financial contract. (E.g. 5 year multi-million dollar contract to mount Google cameras.)

Similar to how private events pay the government for extra police officers to direct traffic or supplement security. Or how the Pentagon charged the Top Gun film production $25,000 to turn an aircraft carrier around.[1] Government entities have set the precedent that they will sell services to private companies.

The citizens' access to the USPS photos would a given and be part of the contract.

[1] https://www.quora.com/Why-did-it-cost-25-000-to-turn-the-shi...

I don't think those two examples are very good for your argument. One is about public safety and the other is basically about military recruitment. There is a lot more precedence for those sorts of things than there is for mounting Google cameras to USPS trucks.

Well, there's a precedence for the government wanting to preserve jobs. USPS lost $5.6 billion last year.[1] A rich Google contract wouldn't solve all their financial problems but conceivably, USPS would hire extra workers to help manage the camera equipment, the photos, and interface with Google Inc.

If not that reason, the government could come up with other justifications. (The US government is very good at coming up with justifications for anything.)

In any case, getting some extra revenue from trucks already burning up huge amounts of fossil fuel to deliver mail seems like an eco-green win/win.

[1] https://about.usps.com/news/national-releases/2016/pr16_092....

the post office isnt the government - its owned by the government, but is not part of the government.

USPS didn't 'lose' money. They were railroaded into paying for pensions in an attempt to kill them off. http://m.govexec.com/management/2017/09/usps-defaults-billio...

The postal service owns tons of valuable land and also could easily charge amazon, etc. competitive rates when abused for their last mile services.

I'd rather my postal service not be treated like another shitty wall street commodity. We all benefit from its continued existence, warts and all.

I am not sure about not losing money.

Please see https://about.usps.com/news/national-releases/2017/pr17_007....

Summary:

- operation cost reduced

- holiday sales saw growth

- first-class mail saw continued loss

It is known USPS has been losing money for many years now due to the raise of emails and now social media and instant messaging.

This is a strong reason some people proposed to privatize USPS instead.

Amazon pays the USPS to roll trucks on Sundays for exclusive (Amazon-only) package delivery service.

So there is precedent for a private company paying a pseudo-USG entity for private benefit.

The Post Office isn't a publicly funded entity - it self funds thru user fees.

In Denmark, several municipalities have hired garbage-collection trucks to test mobile coverage (danish article with images at https://www.dr.dk/nyheder/regionale/oestjylland/skraldebiler...) and some utility companies have done the same with recievers for short-range transmitters placed in water/power/heating meters.

It's a good idea, but they can't have their cake and eat it too. If it's so good for society that they get taxpayer money, the data should be freely licensed. Conversely, if they expect to monetize the data, then we should charge them a very stiff fee for having the monopoly on postal-van cameras (I find it unlikely the postal service would agree to let multiple companies mount equipment on their vehicles, if they'd even let one).

You have it backwards. The postal service mounts their own cameras on the vans and then sells that data to companies wanting to build mapping applications.

That data would then be produced by a US government sponsored entity, and should be licensed in the public domain.

Cost per truck is greater than the value of updating the roads every day. However, having one truck per post office that gets rotated through the routes might be worth it, or even better a few trucks that get rotated through each post office. But, even then the post office avoids a lot of roads in the US.

Security paranoia aside that would be a great way to keep Google Maps up to date even if it was just done quarterly. And frankly I'm sure the US Post Office could use the revenue from whatever Google would pay to do the mapping.

One issue is in the rural areas they delivery mail in their own cars, not USPS mail trucks. But ultimately Google could loan/lease their cars to those drivers.

I'm sure there would be pretty significant concerns about government surveillance were the USPS to be recording every street every day.

I'm sure that would go over really well with the gun-nut/infowars crowd.

UPS trucks instead? Maybe less reach but still pretty good.

> the USPS to be recording every street every day.

Oh, I wasn't thinking that every street every day would get recorded. I don't think Google (or NSA) datacenters have enough storage to keep up with that volume of nationwide photos. (I haven't done a back-of-the-envelope calculation, so maybe I'm wrong and they actually can).

I was just thinking that there are many streetviews (especially rural areas) in Google Maps captured in 2007. My street is from 2013. Using another entity's expansive fleet of vehicles, those maps could be updated more frequently... like once a month instead of every 4 to 10 years. Also, the USPS trucks are already burning the gas driving all those roads so it's a waste of fuel to have another car drive the same route just for photos.

As for fears of surveillance, it's going to depend on:

(1) if the citizens feel they also get benefit from the updated data. E.g. web surfers on Zillow see a house for sale and they would prefer to see a streetview of the neighborhood from last month instead of 5 years ago. Same for scoping out a company's address for a job interview.

(2) Are USPS cameras perceived as a minor extension of traffic cameras at every extension that citizens drive through every day and the police car dash cams that record a continuous loop as they patrol the city? Or it's a major step change in data collection that the public would reject. I don't know.

> I don't think Google (or NSA) datacenters have enough storage to keep up with that volume of nationwide photos.

If you keep taking pictures of the same things over and over, you get a really high compression ratio. Not to mention that you have the option to throw out the old data.

Google datacenters most certainly have enough storage. I'd bet youtube uploads more data / day than these USPS cars would.

Don't forget that the cameras on the cars would not need to have 30 fps uploaded at 4k resolution.

Here's my back-of-envelope calculation:

Assumptions...

  4,000,000 miles of roads 
  (4.12 million minus 164,000)[1][2]

  5280 ft per mile

  200 ft as guess of distance between snapshots

  60,000 bytes as estimate for 1 jpg sized for "near HD"[3]

  15 photos per set [3]

... total is ~95 TB per day.[4] The Youtube stat from 2015 quoted 24 TB per day.[5] The total Google Maps in 2012 size was 20 PB.[6]

So, 95 TB is within the realm of possiblity but it seems like coordinating the logistics in the USA to upload that much from trucks every day would be a huge hurdle. (The Youtube 24TB is worldwide uploads and not just USA.) Since the vast majority of the streetview images for a particular day won't be requested by anyone, it wouldn't be a good return-on-investment to pay for that much disk space.

Obviously, the total goes up if I underestimated the intervals (100 ft instead of 200 ft) or underestimated the jpg size (new generation of HD cameras output 200000 bytes for still images). The total goes down if a significant chunk of those 4 million miles don't need streetviews. There are probably other factors I'm missing. (What's that famous Google interview question again? "How many gas stations are there in the USA?"[7])

[1] https://www.google.com/search?q=miles+of+roads+in+usa

[2] https://www.google.com/search?q=how+many+miles+of+highways+i...

[3] https://petapixel.com/2012/10/15/a-glimpse-of-googles-fleet-...

[4] https://www.wolframalpha.com/input/?i=((4*10%5E6)*5280%2F200...

[5] https://www.quora.com/How-many-terabytes-of-storage-does-You...

[6] https://petapixel.com/2012/06/06/google-street-view-has-snap...

[7] https://www.google.com/search?q=interview+question+how+many+...

95 TB a day would cost Google roughly a million dollars a year. For a company that routinely makes tens of billions, it's not a gigantic amount, and I'm not so sure they wouldn't make it back in extra views - "nearly live" Street View would be an incredible tool. But they wouldn't even have to save all the data for that - just keep it flowing.

As to the logistics of data transfer - why not move the data physically? You're already carrying it around in all the mail trucks. Most of the logistics are already taken care of by the existing postal infrastructure. Just take the SD card out at the depo and put it on the next mail vehicle going towards Google.

Or, to paraphrase Andy Tanenbaum, "Never underestimate the bandwidth of a postal truck with a shoebox of microSD cards hurtling down the highway".

*Fun fact - a shoebox full of 128 gigabyte microSDs has a capacity of nearly 9 petabytes!

Let consumers/business request the update on a geospatial grid square, and pay some amount that could in turn be profit shared with USPS.

ie: a county could pay for its counties streets to save man hours for sqft appraisals, with just change detection, for the annual property appraisal updates.

I investigated targeting individual Uber drivers for street view mapping. Problem is that Uber drivers circle around the same area a lot, so it is not optimal to use professional 360 set. On the other hand, there is no good consumer grade camera for 360 street view mapping yet, although it keeps getting better.