Google, Not the Government, Is Building the Future
nytimes.com
nytimes.com
I was in the Defense Innovation Board small group session with Mattis, Schmidt etc... about a month ago and it was clear that since then the message was received by some but not all (which is partly why I was invited as a reserve officer).
As a result the Deputy Secretary of Defense issued a memo [1] recently which took some of the first steps to get the ball rolling, but frankly it may be too little too late.
OpenAI is generally on the right track, but even then it's primarily dominated by the big players and they don't have incentives to push their best research, data, models, resources to that - of course why would they?
[1]https://www.scribd.com/document/346681336/Establishment-of-t...
Remember the internet, gps, modern telecom etc all came from government research, or some combination of public and private money but were generally declassified 10-15 years after its invention and use (some exceptions obviously). It stands to reason the next generation of AI is already being built in a similar fashion and we'll look back in time and see the patterns in hindsight.
For example, how does the NSA sift through its giant treasure trove of information? Isn't that what AI/ML/Deep Learning is designed to do? In my opinion, the government is very up to date with current trends, maybe more so than people realize.
It's true that a lot of the funding was from ARPA, but I believe that the vast majority of this research was published in standard academic venues. I know of no results in these fields that were classified at the time of discovery and released only 10-15 years later as you claim (but I'd love to see references if you have them!).
Except I can tell you that it's not at least not at the same pace as AFGAM.
Why do you think the message was not received?
The USG is so big with so many different "rice bowls" that it can't efficiently coordinate between all of them - especially not for future facing projects.
It's true that there are pockets of research and innovation in ML across the USG, specifically the DoD, however it's nowhere near at Google/Facebook scale.
Edit: see below. Especially if you disagree with the premise.
Which I do, so I disagree with your premise.
Put another way, if reputable sources contradict you on matters with great impact, do not neglect to fund their voice on the off chance they are correct. This is the same reason why I give the biotech threat guys a voice even if I disagree with their premise.
Better?
That said, ceding control of critical control systems (SCADA for power, water etc...) needs to be done in the right way to actually ensure that the implementation is better, but that's just good engineering.
The simplest metaphor I always make is that it's the same as the elderly being comfortable dying knowing that their offspring are off living their lives. In this case it's humanity that's dying and AGI is our offspring.
I mean we're doomed as a species in the long run anyway - that's well understood evolutionary biology - so lets take control of that and engineer our offspring directly.
Here's Lawrence Krauss making the case in a roundabout way:
"What if our descendants aren't carbon based, what's wrong with that?"
Government should be building the infrastructure of the future.
For example, Google Fiber may have been unprofitable, but the government should be laying fiber to every corner of this country.
Sadly, with 1/3 of the US stuck in propagandaland, this is the best we're going to do.
Money is better spent on anything, but something like low carbon energy would be a much better use of it.
Really, we should focus on getting any broadband to areas that don't have any. Rather than upgrading people with cable internet to gigabit.
1) The benefit is to businesses first, not necessarily consumers. Solid telepresence would allow specialists in various fields (medicine, engineering, etc.) to be located away from the big cities. It would also allow businesses to locate away from the big cities.
2) Part of the reason we don't have new applications is that so few people have gigabit (especially upload). Why write an application that requires gigabit when nobody will use it?
Every order of magnitude jump in communication upload speed spawned something new in technology usage that we couldn't see at the time.
300 baud -> email, BBS
9.6K baud -> graphical Internet
128K baud -> napster, MP3
1.0M baud -> video
So, what is waiting at 10M (we're just getting here in the rural areas and it's patchy), 100M, or 1G upload?
VR!
Stuff like telepresence doesn't require anywhere near gigabit.
I'm sure eventually there will be a need, but it's not today.
When the government invests in something to make access universal, it tends to be a good investment for the country.
When the government interferes in markets, it tends to be a bad investment for the country.
However, now I am questioning that perspective and wondering should the government be shaping our future and if so in what ways?
(On the other hand, Space X versus the usual defense contractors is quite a contrast.)
The role of government, when it comes to minority rights, can act like a big fat rock under the wagon wheel though. As well it should.
After all, when it's not seized by a spasm of science denialism, the government is a chief backer of basic research, which is then often licensed on extremely generous terms to make a wide range of industries and products viable. One could argue that the government is, in fact, uniquely positioned to fund truly basic research, for which risk is often too high - and payoff too uncertain or delayed - to be fundable in a corporate environment. Put another way: the government is often, contrary to belief, much less risk-averse than private enterprise when it comes to research of this sort.
For instance, take this Internet thing we're all communicating over. Without R&D efforts by ARPA in conjunction with several prominent universities, or the willingness of CERN to place key Web protocols in the public domain, we might very well have had non-interoperable AOL-style fiefdoms. But what company would, in the 70s, have funded an unproven telecommunications network just to connect a handful of universities?
When a government tries to build infrastructure directly, you often get disastrous outcomes. Australia's so-called National Broadband Network is a recent and woeful example.
The sole counterexample is probably roads, and even then there's an argument that the public administration of roads has many systematic issues that could be blamed on governmental factors.
What you mean by "long view" here? I also don't particularly see why private companies would (in principle) be better than government-run institutions. I mean, in most of the world universities are government-funded institutions, and I think they produce quite a bit of innovation?
The key here is independence from the direct source of funding. There's nothing magical about private enterprise that means it'll be more "innovative". You just need independence[2].
> When a government tries to build infrastructure directly, you often get disastrous outcomes. Australia's so-called National Broadband Network is a recent and woeful example.
I can't claim to know much about Australia, so can you explain what went wrong with NBN?
I can tell you that e.g. Sweden is among the top[1] of the world when it comes to Internet speed, and I believe that was mostly done by government mandate. (Not sure who actually laid the fiber, but that's pretty much beside the point.)
[1] https://en.wikipedia.org/wiki/List_of_countries_by_Internet_...
[2] As in: Your funding is guaranteed for X number of years, regardless of political whims, popular opinion, etc.
Politics, I believe, is a symptom (of a much bigger problem).
In short, We get what we ask for.
There are plenty of reasons for Uncle Sam to be covert. Obviously, national security comes to mind. The other is the Constitution, as well as other things that are supposed to limit the gov's.
If history is any indication, when it comes to Uncle Sam: just because we don't see it doesn't mean it isn't there.
For example, it's great that publicly funded prior research and experts bootstrapped SpaceX's R&D, and then NASA's contracts for low-priority launches helped them iron out the kinks. It was a nice transition from publicly funded risky R&D to a market-driven private industry. On the other hand, NASA's "full-stack" SLS product is an inexcusable clusterfuck.
Corporations are way better at diving deep to create practical, economical products... very very lean. The government is good at running lots of small risky/low-ROI projects, unprofitable but foundational to future private sectors. Tax the past to subsidize the future.
But pro-society government (centrist), with regular people having direct tools to remove officials, with politicians being responsible for their election promises and with no lobbying allowed, with tight anti-corruption laws and yearly referendum in place.
Unfortunately the future tends to hide its money in offshore bank accounts.
Government also has a greater role to play in extremely long-term initiatives that don't make sense in a shareholder funded vehicle. Much of our current society is founded on government funded R&D as Candu mentions.
https://en.wikipedia.org/wiki/Net_neutrality_in_the_United_S...
Here are some reasons why someone might espouse it:
https://hbr.org/2010/08/why-network-neutrality-is-good
and more specific to the current FCC Chairman:
https://www.eff.org/deeplinks/2017/04/fcc-wants-eliminate-ne...
Netflix didn't begin their streaming service until 2006 or 2007, if my memory is correct. They might lobby for it now (of course they do, it's in their best interest), but they have nothing to do with the original idea or policy.
I happen to think that net neutrality makes for pretty good policy, but I'm open to arguments that suggest otherwise. "Netflix supports it, and therefore it's bad" is not one of those arguments.
Source: http://blog.streamingmedia.com/wp-content/uploads/2013/07/20...
[1] https://media.ccc.de/v/33c3-8064-the_transhumanist_paradox
Well, at the end of the month, instead of paying IBM for mainframe time, we are now paying AWS for the cloud time.
Similarly with the others.
Future building always happens through individuals and corporations who challenge status quo. In fact progress is faster when government is "not looking".
Not sure what you mean by 'propagandaland'. I'm glad so many Americans still believe in limited government. This is one of the most important / best features of this country.
Who believes in unlimited government? Can you name a person or group that does? I'd like to ask them what that even means.
He never said anybody believed in unlimited government. There are plenty of people that believe the government should have more power over individuals. i.e. Socialism, Communism, Socialized Healthcare, etc.
At it's core, it the collective imposing it's will upon the individual. If we continue to give power to the collective to impose on the individual, the individual will cease to have freedoms & rights.
Note that things are even more nuanced. Some individuals hold inordinate influence over the apparatus of the collective, which is used to impose rules on other individuals.
That's kind of implied in expressing joy that there are "still" Americans who believe in limited government. There is no in-between, anything that's not limited is unlimited.
> Socialism, Communism, Socialized Healthcare, etc.
Making murder and kidnapping illegal, and so on? If no, where do you draw the line, what makes socialism, communism and socialized healthcare go together, but not, say, copyright?
> At it's core, it the collective imposing it's will upon the individual. If we continue to give power to the collective to impose on the individual, the individual will cease to have freedoms & rights.
First off, "the collective" is just a whole bunch of individuals. Secondly, it's also a way to for people to organize things they can't hack on too small scale, and a way to enforce some minimum social responsibility on each other regardless of level of individual development. Would you be okay with the frail, young, or otherwise defenseless who can't afford private security and don't have family being free game? I mean, "the government" is what makes it illegal to just steal from and kill poor or demented people without relatives for example; there is NO money in preventing that, the "free market" will not provide for that. Humans do, in spite of it, on the side.
> Some individuals hold inordinate influence over the apparatus of the collective, which is used to impose rules on other individuals.
True. I'm not saying it's not broken, but IMO it's better to repair the ship and discuss modifications than just jump off. There's no land there. I'm not even saying we need a government per se, but we need a society, some minimums, some standards of decency in which the profit for fun and profit can be embedded in, rather than pretending fun and profit can provide the necessary amount of glue. Whether the rules are enforced with cops or remembering the eyes of ones father or however is kind of secondary, really. But capital creating more capital, all by itself, is just an eating mechanism, it's like fire. Not even enough to define the difference between a camp fire and a candle or anything, just the raw process of burning material using oxygen.
Kind of like a power drunk mob, I'll grant that, and I'm not saying there's not plenty of worrying tendencies in that direction, too. I just don't see one as the antidote of the other, I see both more as scapegoats for each other, and ultimately both as undermining the freedom they espouse. We only are as free as the most defenseless members of society are. Well, I am. So in a way, people who say "fuck you, got mine" are directly imposing on my freedom, that they are unaware of that doesn't make it less real to me. Quite the conundrum huh.
Spending $600B/yr on the military is right wing welfare.
Additionally, when massive economies of scale drive the market towards a single provider, your options are to either trust a corporation to behave ethically (given history, a terrible choice) or to run a state monopoly.
There are definitely times when a planned system backed by the state can do a better job of serving the public good than a free market. The right answer is to be flexible and examine each situation independently rather than be dogmatic about government vs markets.
That's how we got into the situation we're in now with respect to net neutrality. Sure there are fixes that involve regulating the private infrastructure, but what's so great about that vs. just making the infrastructure public in the first place?
You can always contract out the building and management of the infrastructure while still maintaining public ownership and control.
That's true whether it's for ill or for good. Last time I checked Google isn't bombing kids in other countries, our government is.
It could only be built with the use of imminent domain.
>Last time I checked Google isn't bombing kids in other countries, our government is.
And at the same time the reason Google (or some other multi-billion dollar company) isn't doing the equivalent of bombing kids in other countries is because the existence of a government with a monopoly on the user of force.
On HN that probably means "Trump supporter".
The left still thinks patronizing the right as "stupid" and incapable is going to get them somewhere.
We detached this subthread from https://news.ycombinator.com/item?id=14362943 and marked it off-topic.