A university IT guy plans to bid in two government airwave auctions | Hacker News Reader

A university IT guy plans to bid in two government airwave auctions

fortune.com

96 pointsapress45 comments

I've always been curious about the mechanics of the FCC frequency spectrum auctions. I know you can view the results on the FCC site below but I'm not sure how to bid and view ongoing auctions: https://www.fcc.gov/round-results-auctions-held-july-2005-pr...

Anyone care to shed some light?

> Anyone care to shed some light?

I'm honestly not trying to be flip, but as far as I've figured out the process goes 1) Find a Law firm on K street that specializes in this. 2) Pay them.

The K street law firm would pull up the FCC’s application instructions: https://docs.fcc.gov/public/attachments/FCC-18-109A1.pdf.

Yes, it’s complex. But you’re talking about auctioning an important public resource for a highly complex purpose. It’s like a permit for drilling for oil on government land. But the process is transparent.

It seems Beyerle (our university IT guy) was able to acquire a local license for a seemingly fair $12k after getting a discount from the FCC no less.

"About 20 years ago, Beyerle was the winning bidder for one of 14 licenses in a high frequency band for his local area, paying just under $12,000 (after getting a 35% discount that the Federal Communications offered to smaller bidders)."

I'm not familiar w/ the FCC process but I was involved in acquiring frequency in a non-US country w/ a non-US company.

The process is elaborate game theory. The rules are simple, e.g.: 3 Rounds, must bid twice, can't collude; 5 spectrums up for bid, 4 bidders; can win 2 only, on success must pay and hit milestones A, B, C by date X, Y, Z.

Often the system is passively rigged - no actual collusion but everyone knows through deductive reasoning what the others will do; and it goes down as expected.

Spectrum auctions have a really neat theoretical history. They're rooted in a 1959 paper by one of the pioneers of regulatory economics, Ronald Coase: https://www.chapman.edu/ESI/wp/Porter-Smith-Hazlett-RadioSpe.... Today, spectrum auctions are the go-to spectrum allocation tool in not just the U.S., but most of Europe as well.

Why auction perpetual rights and not lease them? You could also add a use-it-or-lose-it to avoid dog-in-the-manger exclusion of rivals.

"selling" the rights seems short sighted, no matter how you set the price.

As far as I can tell, the rights are not perpetual. For example, I arbitrarily picked a recent auction for "FM Broadcast" rights (https://www.fcc.gov/auction/98/factsheet), and the description states that the rights being auctioned are for an eight-year license period.

Although it's a bit patchwork, most spectrum rights (from auction to licensure) are use-it-or-lose-it. On licenses as simple as an FM broadcast license, stations must report to the FCC if they will not be broadcasting for any period of time, and if they remain "dark" for too long they face cancellation of their license for non-use. On spectrum auctions it's usually a condition of the license that it will be cancelled if certain conditions specific to the spectrum aren't met, e.g. offering a certain range of services to a certain volume of customers.

Ideally, the government is not in the business of deciding what each block of spectrum should be used for. What Coase proved is that if you auction the spectrum and allow secondary-market transactions in the spectrum, then the spectrum will end up in the highest value use--i.e. the most economically efficient use.[1] It doesn't matter "how you set the price" initially--secondary market transactions will take care of reallocating the spectrum to the best uses.

The FCC auctions don't implement fully Coasian auctions--they auction licenses for a limited duration.

[1] https://en.wikipedia.org/wiki/Coase_theorem

They might end up with their best use, but the public (well ok, government...) would not be compensated for the full value of the public good, right?

e.g. If you initially sell for $1 to BobTelco then BobTelco (and any future “owners”) eventually capture the real value of that spectrum.

It's a pattern repeated across just about every market system: the reward for finding out about a mispricing is equal to the magnitude of the error.

The public can participate in the upside if they buy shares in the public companies that operate their businesss on the spectrum.

But that isn't democratic anymore. Whereas one person equals one vote in the case of the public receiving the profits, participation in the stock market means more profit the more you wealth you already have (for buying shares).

Yes the point is to ensure that resources sre allocated efficiently, not use the resource as a revenue generation tool for the government.

One point: Coase did not write the theorem you're attributing to him -- Stigler did, by cherry-picking the anti-regulatory pieces of Coase's 1960 paper while importantly ignoring Coase's own warnings about the conditions Coase lays out for the case against regulation. These are, and I quote from Coase: "very unrealistic assumptions". The U of C crew nonetheless elevated Coase's conjecture to the rank of a general theory, and a tsunami of right-wing economics quackery has followed this parlor trick.

I suspect there is no honest reading of Coase that supports the notion that the public benefit of the spectrum must be thrown to the market. Readings of the market-fundamentalist Chicago Boys who ran with the favored pieces of his work, sure, but not of him.

  Why auction perpetual rights and not lease them?

Depends if the buyer has a means (and incentive) to clean up the spectrum when their license ends.

Some applications like TV and cell phones make cleaning up the spectrum easily possible - you just turn off the broadcast towers.

But if the government leases the spectrum to a company who uses it for consumer products like wireless doorbells and thermostats and burglar alarms and toy drones and two-way radios, those things are going to keep using the spectrum for years after their license expires.

If im not mistaken, Ronald coase hated this, saw it as a misunderstanding of his work.

His premise was that _in the absence of transaction costs_ it doesn't really matter who gets (for example) spectrum rights. The market will get the rights to whoever can use them best through open market transactions. His conclusion is that transaction costs in reality are high, and therefore it does matter.

A practical issue in the US is that aggregating spectrum is almost impossible, which meant that allocation of spectrum to its current highest value use is dry difficult. The market (because of high transaction costs, in coasian terms) is rigid and inflexible.

I work in fixed wireless and I applaud this guys effort. In my area, Sprint purchased all of our 2.5 Ghz licenses and we had to resort to using the new CBRS band alongside unlicensed frequencies.

Such high frequencies don't seem terribly useful. Aren't they essentially line-of-sight?

I'd imagine that explains some of the lack of interest.

Yep. WiGig is 60GHz and won't penetrate even a dry wall.

It would make more sense to allocate these frequencies for personal use (although WiGig has that covered), or technologies similar to Artemis pCell[1] (possibly vaporware at this point).

[1]: https://www.artemis.com/pcell

Even 5GHz wireless phone base units won't penetrate a single layer of standard concrete blocks, like you have in home construction in Florida. Yet 2.5Ghz (WiFi etc.) or below (presumably part 15 at 100 mw) exhibits very little apparent attenuation throughout or outside the house. You may want to check out the Atmospheric Absorption by Frequency chart at the RFCafe site:

http://www.rfcafe.com/references/electrical/atm-absorption.h...

Would it be possible to quantum tunnel the wave packet through the solid wall of concrete? (In this case it would be the wave produced by an RF device)

To experiment with quantum tunneling you're dealing with phenomena manifesting at the 1nm scale. One macro level theoretical app for quantum conductivity seems to involve wave packets interacting with perfectly rectangular uniform tightly-spaced cation-nuclei barriers which theoretically allows free electron tunneling (IE, energy transmission). [1] Even if it works, seems like a lot of trouble given that an array of duplex repeaters in the attic could solve the problem.

[1] https://en.wikipedia.org/wiki/Quantum_tunnelling

BTW Anki as in J-flashcards or as in Robotics?

> One macro level theoretical app for quantum conductivity seems to involve wave packets interacting with perfectly rectangular uniform tightly-spaced cation-nuclei barriers which theoretically allows free electron tunneling (IE, energy transmission).

That’s an interesting use case for quantum tunneling. 1 nm scale is hard for me to visualize mentally. When dealing with quantum mechanics, it seems that one should not even attempt to use intuition. The way systems behave at quantum scale is completely counterintuitive to me. Do you have any recommendations for lectures or books?

> BTW Anki as in J-flashcards or as in Robotics?

Robotics.

A) math refresher: Basic Training in Mathematics - Shankar

B) undergrad A-level text: Principles of Quantum Mechanics - Shankar

Happened to like his approach given that he's trying to convey information with serviceable clarity given the subject matter without trying to impress you with his divine intellect, like some of the pedagogues who are legends in their own minds

C) Road to Reality - A Complete Guide to the Laws of the Universe - Penrose

Last one sounds pretentious but was a magnum opus. His postulates concerning consciousness as a synthesis of quantum functioning were probably a bit much.

FWIW when studying Japanese back in 2002 I liked the Heisig method but wanted animated stroke order displays for the characters. Jim Breen at Monash in Australia had half the set in .gif but I had to coax Jack Halpern in Japan to release the other half. Armed with the full set I was finally able to digitize the Remembering the Kanji in a useful .html form for personal use over the space of a month (still keep it as a bookmarked local app in Firefox). When done, apps like Anki (tr: memorization) J-flashcards started showing up, duplicating the work. :)

No, that is not how quantum tunneling works.

How long do air wave licenses last? Is there a substantial cost to renew or do you simply have to re-bid?

I honestly don't know why they are focusing on towers. I honestly believe we need to skip that and use LEO sats. While more expensive upfront, it also provides the ability to give global coverage. Some googling will show you white papers with estimates of ~4500 nanosats (30x10x10cm), but you can reduce the number of those by not rolling out usage worldwide and then scale up (entering markets in other countries. Even ones that have NO internet access currently).

You have low latency, high speeds, and don't need as many towers. But you HAVE to be in LEO.

Because of how orbits work, I don't think you can really "focus" on an area by reducing the number of satellites without a corresponding decrease in availability and/or throughput. If you have a constellation that has consistent coverage over the US, you're also going to have simultaneous coverage over a good chunk of the rest of the world (although possibly not anywhere more northern than the United States).

You do realize that I addressed this as actually a feature, right?

You don't need 4500 to provide global coverage and handle 100m users. And besides, this isn't about orbits, that is about load balancing. Orbits alone doesn't even describe your coverage area. They do provide an upper bound on your viewable area, but you're never going to reach that.

https://en.wikipedia.org/wiki/Geostationary_orbit

Geostationary orbits are very far away. The speed of light being what it is, they aren't great at relaying realtime two-way communication.

GEO is also significantly further than LEO.

Give it a decade and there should be one or two to try out.

[1]: https://en.wikipedia.org/wiki/Starlink_(satellite_constellat... [2]: https://en.m.wikipedia.org/wiki/OneWeb_satellite_constellati...

I'm fully aware of these programs, and a few more (I very recently left the space industry). I'm more concerned that these initiatives are not being pushed forward faster. As well as these things seem to focus on "We can provide the world with internet" and not "We can provide America with 5G and continually scale to include substantially larger markets". I think that's why they aren't getting funded. Because they are stating the benefits of the grand design when companies like Verizon and AT&T are just focused on the next step.

Backhaul

A ViaSat-3 costs about $600 million and offers 1 terabit of capacity. That's about as much as 50 5G cells, which you can build for about $150,000 each (or about $7.5 million to support the same capacity).

ViaSat-3 is 6400kg (Aka, a large sat) and placed in GEO. It is also using rad hard hardware and has significant redundancy. Comparing that to cube sats in LEO is like comparing apples and oranges. Small sats are... well small and generally equipped with COTS devices, which greatly drops the price. Also there is a significantly different cost to putting something in LEO vs GEO. (GEO costs several times more per kg)

Is there a cube sat with COTS hardware that has a 1 tbps capacity? Otherwise, you’re comparing real systems that are being deployed next year with paperware.

Online I’m seeing entire constellations of 300-600 small satellites weighing in at 5-10 terabits of capacity. You need about 500 5G towers to get that, which will cost you about the same as a single Falcon 9 Heavy launch. Can you launch an entire constellation of cube sats for $75 million? (Far less with 5G small cells.)

Also, what does the mobility story look like? You’d be a fool to spend billions on fixed-only infrastructure in 2018.

> Is there a cube sat with COTS hardware that has a 1 tbps capacity?

> Online I’m seeing entire constellations of 300-600 small satellites weighing in at 5-10 terabits of capacity.

>>> Some googling will show you white papers with estimates of ~4500 nanosats (30x10x10cm), but you can reduce the number of those by not rolling out usage worldwide and then scale up

I'm sorry, but it feels as if you're willfully ignoring a VERY important part of my comment. Yes, you need a large number. But you don't have to start at the 4.5k constellation estimate. I'm also not saying it is strictly cheaper, but I am saying there are some lucrative advantages to it. That's what I'm getting at. But I guess the responses I'm getting answer my original question.

I’m not ignoring that, I’m asking how the economics works. 5G will involve deployment of tens of thousands of cells in the US alone. Even 4,500 nano sats won’t offer similar capacity unless each sat has massive capacity. Moreover, mobility is crucial. If you can’t put the reciever in a cell phone, the utility is greatly diminished. Unless my math is wrong, it sounds like the answer to your question is: people are investing in 5G rather than sats because it offers more capacity and a dramatically more valuable (mobile) service. The benefit of being able to cover the world with internet isn’t important to the FCC or US cellular companies.

Well here's something to get you started. One of those whitepapers I told you that you can easily google. Mind you, none of them is exactly complete (once they are, they stay internal to the company).

https://arxiv.org/pdf/1508.02383.pdf

> How A University IT Guy Looks To Outmaneuver Verizon and AT&T In Upcoming Spectrum Auctions

By buying spectrum in an area where they don't want to buy, it turns out.

Fun fact: I recently outmaneuvered AT&T in produce sales, by buying some tomatoes they weren't shopping for at the grocery store.

(Also, given that Time Warner rebranded itself as "Spectrum", this is not what I thought the article was about; can we lower case the headline except for proper nouns?)

We've replaced the title with more representative language from the article.

Reply on news.ycombinator.com