Alphabet's Project Loon says it will run the program with fewer balloons | Hacker News Reader

Alphabet's Project Loon says it will run the program with fewer balloons

bloomberg.com

124 pointsJumpCrisscross60 comments

So the story after the headline is that Alphabet hired a seasoned Satellite Broadband guy, Tom Moore, to run the division and he pushed them to make the project more efficient, and ultimately profitable.

This has Ruth Porat's stamp all over it - they hired her as CFO to bring "financial discipline" to Alphabet, and among those duties was to rein the moonshot stuff into actual businesses, or kill them off.

I actually think this is good news for Alphabet's long term ability to innovate: if they can demonstrate that some of these Moonshots can stand on their own their investors / board will be more apt to provide for freedom to innovate over the long term.

Or they'll just stop doing them.

And someone else will fill the void. If there's one thing I'm not afraid of scarcity, it's human innovation.

I, however, am afraid of scarcity in resources for that innovation. Moonshot projects don't make business sense, so when beancounters notice, they happily axe them. And VCs are also focusing on short-term returns, hence the proliferation of bullshit one-trick-pony startups that don't create anything of much value. Moonshots are way underinvested, IMO.

Slowing down Google Fiber doesn't sound good to me, while it probably comes from the same idea.

I think the idea behind the GFiber shrinkage is that WebPass achieves the same end result, but doesn't require the extremely time-consuming, expensive, and risky negotiation for right-of-ways on the poles or digging permits that GFiber would've. If so, then economically at least Webpass is the way to go. At least, that's the theory. It'd suck to scale back GFiber and then find out that Webpass wouldn't actually scale to a nationwide roll-out either.

I'm sad to see GFiber go as well, I worked on it right before its launch and was really hoping I'd get to try it out as a customer.

Wireless does not achieve the same result. It has a lot of downsides and more limited application.

I think that is political.

What kind of politics do think it is?

The kind that grants a monopoly to ISPs which allows Avg. Americans to have 1-2 choices for internet >10mbs. There was a 1 Trillion or greater global roll up in the last 2ish years.

I think that google was likely meeting friction somewhere in the process, prob everywhere. I am sure they also got pushback internally but I figure launching in major cities w/ 10-20 year better cable would be profitable. Especially since that industry is insanely profitable

I thought their whole point was to break through such crony politics, to prove the point. But I guess they don't have the guts for it.

If many years ago, someone told me that a humongous company named Alphabet was thinking about deploying balloons all over the world, I'd have told you a thing or two about having a charming imagination.

This is an absolutely horrible and misleading headline. I was stunned when I read it. I'd just read earlier in the day how the Google scientists had ingeniously figured out how to keep balloons in a single spot for months at a time greatly reducing the projects cost.

So basically Project Loon is still going to blanket the globe with Internet balloons. They're just going to accomplish the same original goal with far fewer balloons.

The article seems to indicate that they're downsizing that project to a small experiment, with 10-30 balloons. Sort of like Google Fiber, which had huge hype and very modest deployment. Before that, there was Google Public WiFi, which covered Mountain View, CA badly, and then went into some Starbucks. Maybe Google/Alphabet should stop trying to be a carrier. Their track record in that business is terrible.

google is an expert at soaking up the good publicity for their "moonshots" and then not delivering. and they slowly fail to where people don't notice it and get excited for the next shiny thing

The saying is you miss 100% of the (moon)shots you don't attempt.

The point of the parent comment is that they don't fully attempt them. They put forth just enough time and money investment to make a big PR splash and get a proof of concept out the door, then slowly back away from things.

So how do you define "fully attempt"?

Both projects made immense progress over many years -- they just weren't able to justify the costs. IMO, they were actually really good attempts.

How do you define "immense progress"? Fully attempt to me means actually sinking a significant portion of overall company budget (say, greater than 5%) into a project, marketing it, and attempting to spread it to a wide market. Or put it another way, if a venture-backed startup were pursuing one of these avenues, what would they do? Certainly they wouldn't be content with pilot projects that fizzle out and glacially slow deployment across the country. If Uber were a Google moonshot, they would still be running only black cars in SF.

Google is fantastic at highly-automated software, but really sucks at any kind of real world hardware or service.

Inflate the stock price with "moonshots". Slowly kill off the moonshots and inflate the stock price with "fiscal responsibility".

Google Fiber had one good side-effect: motivating AT&T to deploy Gigabit fiber in downtown Mountain View -- it's here and it works.

And I'm hearing from my Loonie friends who got laid off. They don't exactly spin this as a positive development.

Yes, being able to use stationary balloons means that the minimum viable project can be a lot smaller. And now it is.

I'm sorry about your Loonie friends, but I can't help but think that having a real commercially viable deployment could be anything but a good thing for Loon.

The tech is sort of meaningless if it can't be cost competitive with other options.

Totally agree. This seems like a clickbait title to cash in on the "yet-another-cancelled-Alphabet-moonshot". It's unfortunate that this particular article about this story is on the front page of HN when there are several others[1] that contradict this headline and actually even provide more information.

[1] https://techcrunch.com/2017/02/16/how-googles-project-loon-b... http://www.businessinsider.com/project-loon-x-google-alphabe... http://www.recode.net/2017/2/16/14640470/alphabets-balloon-b... https://news.fastcompany.com/alphabet-uses-ai-to-expand-cove... https://www.cnet.com/au/news/google-alphabet-project-loon-la... http://mashable.com/2017/02/16/google-loon-balloon-internet-... https://www.theguardian.com/technology/2017/feb/16/google-pr... http://www.dailymail.co.uk/wires/afp/article-4233028/Alphabe...

We edited the title (https://news.ycombinator.com/item?id=13665941).

The difference in perspective is between people who are buying Google's marketing copy as written, and people who are looking at the trend of how they're handling all of their project failures across the board. Every other Google moonshot is dead or dying, and news about Project Loon going south isn't exactly new out of left field.

> Loon engineers turned to the computing power of Google to improve the navigational system

That earned a chuckle.

Maybe musks satellite idea is better and they like it?

The problem is the speed of light is too slow :( Latency just orbit and back is 50ms+

There are lots of different orbits. The ISS is 250 miles away, which shouldn't cause much of a latency problem. Of course, it's a compromise between the number of satellites and the latency.

There's also issues with how fast the satellites move across the sky. At low orbits they fly by and you either need ridiculously sensitive wide angle antennas or solid-state steerable high-gain antennas that track them as they go by.

See ex: https://www.kymetacorp.com/

It's a solved problem, the question is how expensive the ground stations are.

Iridium phones have always worked with fast-moving satellites. The 9555 phone is $939 on Amazon, and the Iridium Go hotspot is $689. Ideally, an internet access terminal would be closer to $100, but if you're sharing the connection with a bunch of neighbors using mesh WiFi, $689 is doable.

Iridium GO has the bandwidth of a dialup modem.

But there's no reason why higher bandwidth equipment is going to be more (edit: I meant hugely more....) expensive: same pointing problem to solve.

BTW there's going to be higher bandwidth (edit: adding Ka band to the previous L) from Iridium once Iridium NEXT is fully up.

Higher bandwidth means more noise, which means you need either higher gain or more sensitive detection (until you hit thermal limits). They're very interrelated issues.

The overall discussion in this sub-thread is the cost of being able to point at fast-moving satellites.

For straight bandwidth, how the the price of phone modems go up from the first analog phones to 5G? Seems to have not been much of a burden to phone prices. (Edit: shaped beams in the cell towers make a huge difference.)

I'm addressing your point that it's a solved problem. The issues around making high bandwidth data interconnects are not just "straight bandwidth". I'm explaining that there are fundamental interconnections between bandwidth and SNR that make large bandwidths harder to build at sufficient SNR:

https://en.wikipedia.org/wiki/Johnson%E2%80%93Nyquist_noise

The iridium satellite receivers use low-gain antennas, so reception is limited to small bandwidths. This means they don't care where the satellites are. New technology will enable high-gain antennas that can track the motion of satellites, enabling much higher channel rates.

I think you misunderstand what I meant by "it". I added a few notes to my earlier comments; you're assuming that the new satellites don't have higher power or tighter beams? Not to mention that low-end WiFi equipment now has shaped beams, so it's not exactly rocket science to have that in an Iridium receiver. As a former radio astronomer who did some VLBI, solid state phased arrays are easy to understand, I'm mostly glad that I don't have to be the person making them affordable in a few-hundred-$ ground station!

Yeah but aren't they also ancient in internet terms.... weren't they placed up there in the mid 90s?

The trick that most of these folks have gone with is a active phase array antenna. These antennas aren't small (smallest I've seen is 1x.25 meters, but they "just work" -- you leave them outside and suddenly connectivity.

Says they're intended at 1100 kilometers: https://en.wikipedia.org/wiki/SpaceX_satellite_constellation

On a straight shot with no switching time that's up and back in 7.3ms.

This article [0] says they expect latency to be between 25 and 35ms with the 4,425 satellite constellation.

[0] https://arstechnica.com/information-technology/2016/11/space...

That's a poor title considering they've actually had a lot of success with their balloons, so much so that they will be able to use less of them. That means it's actually more likely to happen.

Ok, we edited the title to use language from the article. If someone can suggest a better (more accurate and neutral) title, we can change it again.

By the comments here it looks like 8/10 people on HN just comment without actually reading the article itself. Google aren't scrapping the whole idea of balloons, they just found a better way and faster way to make it a profitable business.

It just looks this way because 2/10 manage to guess what the article is about.

Reddit trained me to comments first then article if it's worth it. I imagine it's the same for a lot of people. Far too often the top comment is telling you all about why the article you were about to read is complete bullshit.

I'm the same but that's because of the so-many bullshit articles I had to read before figuring this out. Now I read the comments first and learned to stay away from certain sources with frequent low quality content. Also, this happens when you give hackers imaginary points; they want to game the system.

Slightly tangential, but I'm really interested at the political influence of what space satellites or balloon Internet might do.

I'm considering a scenario where a nefarious, space-capable entity is capable of disabling satellites. Idk I guess from the perhspective of like a civil engineer or something, it seems that an idea where your infrastructure is out of reach or particularly vulnerable seems like a bad idea.

There could be """accidental""" space object collisions, direct attacks on them, or even just unforeseeable maintenance difficulties trying to take care of dozens or hundreds or thousands of objects not immediately available to us, serving an incredibly critical and global backbone of infrastructure, for more or less every industry on earth, or potentially ventures off earth as well.

What an exciting time to be alive, it's like some Twilight Zone Star Trek mashup lately.

Edit: the title of this post is pretty dumb btw, I contend it should be changed to reflect changes in the program, not its abandonment

> it seems that an idea where your infrastructure is out of reach or particularly vulnerable seems like a bad idea.

Is it better to have your infrastructure where someone with a shovel, or a fishing boats dragging its anchor can cause catastrophic damage?

No, but you might notice I didn't exactly advocate for under sea cable proliferation either. It has the same (percieved, I'm no expert) flaws.

So how should we connect e.g. the US to Europe, then? Sure there are flaws and vulnerabilities, but the alternative seems ... worse.

Well I don't know, I didn't say I knew. I'm just pointing out that there are obvious weak points in critical infrastructure in space, as there are weak points in undersea cables, etc. Lots of infrastructure has bad aspects. I wish I had some deeper, enlightening insight about it all, but I don't.

I thought this was a cool project

Reading the article is helpful to make informed comments.

Who does what with what?

I would think it would be demoralizing, over time, to work on ambitious projects that turn out to never be fielded. (See also the linked stories on Titan and Wing, e.g. https://www.bloomberg.com/news/articles/2017-01-11/alphabet-...) One of the draws of working for a bigcorp is that your work has the power of the organization behind it, to help it succeed.

...except that's not what it's saying at all. They're scaling back their initial goal to deploy worldwide because it's working so well in smaller deployments, and the worldwide deployment doesn't make sense.

Strong disagree. Every other sentence in the article is about scaling back! Which successful small developments did you see substantiated in the article? The Sri Lanka development mentioned in TFA seems to have been abandoned (http://www.dailymirror.lk/article/-Proposed-Google-Loon-proj...). The one in India seems to have been abandoned.

It looks like the project has narrowed down into a focused "small demo" phase that is no longer commensurate with the original scope of the project.

Remember how Google exited Fiber? It was the same combination of re-scoping and backing away from nationwide ambition, over years.

It sucks when novel, ambitious engineering ideas don't get the support initially promised (I could tell stories...), but we can't fail to see the directional information given in the article.

The news about Sri Lanka is disappointing, but unless you think it's entirely spin it seems reasonable to look at other countries where carriers may be friendlier and spectrum easier to acquire.

Reply on news.ycombinator.com