HNHacker News
TopNewBestAskShowJobs

blamestross

696 karma · joined October 16, 2016

Mad Computer Scientist https://miasma.space

distributed hash tables, p2p systems, supply chain and network security.

submissionscomments
blamestross··on Recovering from AI addiction
You are not alone! I used programming flow state as pain management for a decade.

Its one of those "paid for your mental disorder" situations that are a lot more common than people realize.

blamestross··on Ask HN: If AGI were invented tomorrow which countries would fare better?
Its always funny seeing AI safety discussed and pointing at it as "that is just the problems with capitalism in microcosm"
blamestross··on Ask HN: If AGI were invented tomorrow which countries would fare better?
You live in a world with MANY AGIs.

Collective and Swarm Superintelligence isn't a new thing at all. We call them companies, governments, organizations and churches. They just (mostly) run on meat and memes.

The only recent change is that a lot of them are dangerously powerful and paperclip optimized to produce "shareholder value".

I know "capitalism is a runaway swarm superintelligence" isn't the scifi future you want, but what criteria is it missing? The curve isn't the exponential kurzweil dreamed of, but that has only ever been a marketing pitch, all growth curves are punctuated sigmoids.

As to who does well? Who has aligned the optimization criteria for thier Super-AGI with their actual well-being? Who hasn't?

blamestross··on Game publishers respond to Stop Killing Games claim it curtails developer choice
> People would have the same expectations as now

If that premise was true, why would misleading advertisement be the norm right now? Why bother?

Changing it to a subscription WOULD change perception. Most people don't understand the current status quo. When they do know, it would create a market pressure for real game ownership.

blamestross··on Stop Killing Games
Ross talks about disclaimers and honesty in advertising being reasonable solutions all the time. He kinda is the "Movement".
blamestross··on Stop Killing Games
Right the problem is that they didn't do that. 3 months notice isn't the same as "making it clear to the customer at purchase time how long this product might reasonably be supported"

I'd certainly encourage a more sustainable solution like a self runnable server, but I'd settle for replacing the word "buy" in marketing with "license" or "rent" with actual terms other than "until we decide to turn it off".

You don't get to make software a "license" but then not have any obligations to your licensee.

blamestross··on Stop Killing Games
Its been one guy and a discord working nonstop for over a year. Not exactly the situation you imagine it.

Its also been overwhelming ignored and not-viable in most countries. The US for example is a lost cause.

If you think this is things working out well for middle class folks, then similar work for human rights violations are utterly doomed.

blamestross··on Stop Killing Games
You could also comply by providing a self hostable server, or even source code that might be convertible to a self hostable server.

Requiring an "end of life plan" in the ToS would be a start.

blamestross··on The Consciousness Gradient: When Machines Begin to Wonder
You are right, i should have added "like I do" to that statement.

Its about the fact i shouldn't be using my internal model as the basis for my expectations of their behavior. I need to suppress that instinct and learn a more correct model of what an animal perceives and how it reacts that isn't built on anthropomorphic projection.

Based on the response, I need to try communicating this entire sentiment differently.

blamestross··on The Consciousness Gradient: When Machines Begin to Wonder
The cultural fixation on "Consciousness" has become deeply frustrating to me.

We are naturally "animistic" and personifying. The twist in structure allowing mirror neurons being able to re-use the hardware for thinking to model the behavior of external things is useful and has been critical to our success. Unlearning the behavior of of that animism is HARD. Maintaining awareness that the forces of nature, animals, or even objects don't have feelings, motivations, and narratives of their own is hard work but also becomes a more accurate and useful model of reality.

I think that dissonance between hardware that wants to interpret the world as a reflection of self, and the forced acknowledgement that it is not is uncomfortable. And we keep filling that discomfort with whatever rhetoric we can force to fit, and once that schema is in place, it takes a great act of discomfort and bravery to remove or replace it. The arguments and debates about it don't change minds, they just exacerbate the dissonance, making people even more motivated to shout loudly their model of the situation is right.

I desperately want the answers too. I don't know any of the answers. I don't think our culture (or even our neuroanatomy) is ready for the answers. In the meantime people yell at each other a lot without wanting to listen.

blamestross··on Google Unveils On-Device Sign Language Model for Translators and LSPs
My bet is that Dr. Thad Starner's dataset is backing this:

https://sites.cc.gatech.edu/home/thad/031_research.htm

https://www.popsign.org/ is one of his projects and he has worked with them before on google glass.

blamestross··on On memes, mimetic desire, and why it's always that deep
Memes are well labeled. A semantic compression of an entire sentiment. The pinnacle of communication by compressed experiential and cultural context.

And humans like compressed communication, makes them feel warm fuzzy and agentic when they can take in a external-coded token and decompress it into a bunch of internal-coded tokens.

blamestross··on Life as Slime
The nifty feature of intelligence is that you get to examine what you value. Critical peer comments (so far) don't seem to address the "naturalistic fallacy" part, or even lean into it.

As far as I am tell, life can be maximally ambiguously defined as "entropy deferral". Nothing can stop entropy, but life crams as much organization into that lifetime as possible. I think that is kinda cool and I want to help, so I think we should make as much matter alive as possible before the universe fizzles out.

Rationalizing value judgments is always a challenge. We can argue over facts and implication of facts all we want, but the "predicate values" are arbitrary and you can't change them in others. I generally don't bother unless I have a strong idea of my audience's "predicate values". If they don't match mine, or I can't manipulate others into agreeing we share instrumental values, I am just out of luck.

blamestross··on Homomorphically Encrypting CRDTs
> Which means it would need to sending and receiving the entire state of the CRDT with each change. > If the friend is online then sending operations is possible, because they can be decrypted and merged.

Or the user's client can flatten un-acked changes and tell the server to store that instead.

It can just allways flatten until it hears back from a peer.

The entire scenario is over-contrived. I wish they had just shown it off instead of making the lie of a justification.

blamestross··on Google aims to reinvent email with Wave (2009)
I feel like Wave was "we figured out CRDTs lets go nuts with them"

Many of the features that were revolutionary did get baked into docs over the years. I don't think people realize just how similar google docs has become to wave over the years.

The parts missing are the breadth of features and extendability.

blamestross··on How to Build Conscious Machines
Looks a lot like the game theory concept of the "level" of a player. (how many levels of metacognition in others can it practically take into account)
blamestross··on Research suggests Big Bang may have taken place inside a black hole
Cosmic Background Radiation distribution could be that information. The distribution of mass hitting the event horizon then bouncing
blamestross··on Research suggests Big Bang may have taken place inside a black hole
*aparent mass goes up

Things don't get more mass, they just take more energy to accelerate which looks a lot like more mass.

It doesn't imply for example, a high speed mass would cause more gravitational attraction than a slow one.

If that was the case, a black hole would be even worse as it accelerates matter towards itself and it gains "bonus mass"

blamestross··on Research suggests Big Bang may have taken place inside a black hole
The problem is that:

- We have a parent universe we will never be able to observe.

could be a true statement.

The "infinite sequence" part is just a likely implication, it isn't necessarily true. We would need information we can't access to find out.

blamestross··on One Man Armies
Now if only i could afford to spend a few years working on something like this. Sadly there is a mortgage to pay and investment banks who want returns this quarter, not in 3 years.
blamestross··on Chomsky on what ChatGPT is good for (2023)
> Does the process matter in all cases?

So there are 2 dimensions being conflated here:

"Does how the reasoning work matter in all cases" Pretty Obviously no, but it may matter in some of them. We also don't really understand which ones yet.

"Does the reasoning work as intended in all cases?" Pretty Obviously no, but it doesn't work for at least some of them. We also don't really understand which ones yet.

"We also don't really understand which ones yet" Is the critical point of caution.

blamestross··on Chomsky on what ChatGPT is good for (2023)
A computer isn't a math capable machine.

> Perhaps it is more important to know the limitations of tools rather than dismiss their utility entirely due to the existence of limitations.

Well, yes. And "reasoning" is only something LLMs do coincidentally, to their function as sequence continuation engines. Like performing accurate math on rationale numbers, it can happen if you put in a lot of work and accept a LOT of expensive computation. Even then there exists computations that just are not reasonable or feasible.

Reminding folks to dismiss the massive propaganda engine pushing this bubble isn't "dismissing their utility entirely".

These are not reasoning machines. Treating them like they are will get you hurt eventually.

blamestross··on Problems in AI alignment: A scale model
I'm kind of upset to see systematically "Alignment" and "AI Safety" co-opted for "undesirable business outcomes".

These are existential problems, not mild profit blockers. Its almost like the goals of humanity and these companies are misaligned.

blamestross··on How Obama’s BlackBerry got secured (2013)
In terms of technical abilities: - "commercial solutions are equally good or better"

The way governments can have an advantage is in where and how those services are managed. Events like "signalgate" scare me because it means we are not leveraging that effectively due to bad managers.

blamestross··on How Obama’s BlackBerry got secured (2013)
I can only speculate based on publicly available information.

When a was in grad-school, "state level actors" were the boogeyman. You were told to just assume that everything would be compromised to them.

I ended up specializing in p2p systems (distributed hash tables, overlay networks, communication systems) and "State Level Actors" become "in scope" for me. Modern cryptography is focused on the capabilities of hypothetical computers and making radically more computational power required than is reasonable to expect of current human economies. Backdoors in the codebases for encryption is a fun hypothetical, but the level of scrutiny they are under would require a conspiracy beyond any i could imagine to hide.

Eclipse and Sybil attacks were the real threats. Those are Operational attacks, not Signals.

Now that I have spend a decade in the security industry in larger corporations. "State Level Actors" are entirely in the threat model. We don't talk about it explicitly, but these companies stand to loose globally if any one government compromised them. Government funded actors are assumed to be the primary threat. Supply chain attacks like XZ are the ones that scare us, the ones we might have missed. That came from superiority in operations not technical superiority. They actively pay me and a bunch of other people a LOT of money to actively detect and prevent issues like this.

The other side of the argument is a human organizational one. The story of this decade of military spending is outsourcing. Biden's Supply Chain Security EO and the new DoD software procurement requirements are bandaids on gaping wounds.

Even with it's massive defence spending budget, the TLAs couldn't keep up with the industry while also securing all its software. They have 3rd party dependencies too. Assuming that they don't just allways directly outsource.

And why bother? These companies have the entirety of human communications MITMed. Why bother with a complex secret system when a FISA warrant is cheaper and more efficient. PRISM(For attack) and TOR(For defense) stand out as successes of operational attacks. They don't need technological superiority.

I fully expect TLAs maintain an android fork and linux forks, but that is opsec for dependency management, not adding special sauce. The industry simply has more resources and more eyes on the problem than the government could ever afford.

The last part is simple "Brain Drain". The people who are really good at this generally don't want to work for the government and have done too many drugs to ever get clearance. Unless they have a lot of security engineer salaries in classified budgets they also can't afford us. Governments have direct agents that are underpaid and underskilled and they have working relationships with criminal organizations who work deniable offense for them.

Opsec is clearly their leverage-able resource, why not lean into it almost exclusively?

blamestross··on How Obama’s BlackBerry got secured (2013)
I'm increasingly of the belief that modern governments have lost all advantage in the space of security hardware and software. They only have OpSec and a monopoly on violence to leverage in order to have an improved security situation over the public sector.

They don't seem to be using that OpSec superiority effectively right now.

I work public-sector in supply chain security and I am terrified that the situation we currently have in the corporate world is actually the best there is.

blamestross··on What You Shouldn't Know About Quantum Computers
lots of functions with horizontal asymptotes look ok on a log scale until they don't.
blamestross··on What You Shouldn't Know About Quantum Computers
You have to get to page 25 before it starts being honest about the fact quantum computing is a con.

Important Nuance: the research is real, the science is real, but the narrative being sold about the future of quantum computing to ensnare investors is not a reasonable prediction and is a con.

blamestross··on The Nonprofit Industrial Complex and the Corruption of the American City
California also gets at least one countries worth of homeless shipped to it BECAUSE of the resources it provides (And political spite). This is the bad Nash Equilibrium in action, being punished for improving the situation.
blamestross··on Google AI recommends adding Elmer's glue to pizza cheese after scanning Reddit
Bake vs Cook are they key words here.
← PreviousPage 5 of 21Next →