HNHacker News
TopNewBestAskShowJobs

jabowery

27 karma · joined November 9, 2018

A consistent history of leading edge technology development, most recently with the Diogenes Institute's comprehensive plan for energy and the environment based on photosynthetic fixing of CO2, leading to involvement with Algasol's photobioreactor technology.

Other leading contributions started with the first networked virtual reality gaming system (Spasim), the first mass market network information service (Viewtron), the first prize for rocketry leading ultimately to the Ansari X-Prize, the first federated login, authentication and authorization system for web-based customer accounts (HP), the first reliance on top-down universal artificial intelligence (AIXI) for natural language understanding (Hutter Prize), and what may have been (unknown due to secrecy requirements in this area) the first high frequency cryptocurrency arbitrage system.

Public life includes leading commercialization of space launch services, testimony before Congress on privatization of space launch services and the first Ka-band communication satellite license issued by the FCC.

submissionscomments
jabowery··on Bard is getting better at logic and reasoning
The point of this "IQ Test" is to set a relatively low-bar for passing the IQ test question so that even intellectually lazy people can get an intuitive feel for the limitation of Transformer models. This limitation has been pointed out formally by the DeepMind paper "Neural Networks and the Chomsky Hierarchy".

https://arxiv.org/abs/2207.02098

The general principle may be understood in terms of the approximation of Solomonoff Induction by natural intelligence during the activity known as "data driven science" aka "The Unreasonable Effectiveness of Mathematics In the Natural Sciences". Basically, if your learning model is incapable of at least context sensitive grammars in the Chomsky hierarchy, it isn't capable of inducing dynamical algorithmic models of the world. If it can't do that, then it can't model causality and is therefore going to go astray when it comes to understanding what "is" and therefore can't be relied upon when it comes to alignment of what it "ought" to be doing.

PS: You never bothered to say whether the program you provided was from an LLM or from yourself. Why not?

jabowery··on Bard is getting better at logic and reasoning
Yud is doing more than his share of generating misconstrual of his own statements as evidenced by the laws and regulations being enacted by people who are convinced that AGI is upon is.

Ironically, they're right in the sense that the global economy is an unfriendly AGI causing the demographic transition to extinction levels of total fertility rate in exact proportion to the degree it has turned its human components into sterile worker mechanical Turks -- most exemplified by the very people who are misconstruing Yud's statements.

jabowery··on Bard is getting better at logic and reasoning
Wrong output.

What you call "code golf" is the essence of the natural sciences:

Inducing natural laws from the data generated by those natural laws. In this case, the universe to be modeled was generated by:

print(‘’.join([f’{xint:0{5}b}’ for xint in range(32)]))

jabowery··on Bard is getting better at logic and reasoning
It doesn't take much to check the output of that and see it isn't off by a large amount.

As for the answer, look at it in groups of 5 bits.

jabowery··on Bard is getting better at logic and reasoning
The "more complex task in mind" was, of course, to generate the "shortest" program. GPT-4, by asking for a "certain pattern" is attempting to have you do the intellectual heavy lifting for it -- although in this case the intellectual lifting is quite light.
jabowery··on Bard is getting better at logic and reasoning
There are plenty of those who purport AGIs threaten us and conflate "existence" with "potential". This is aimed at those driven to hysterics by such.
jabowery··on Bard is getting better at logic and reasoning
Ask any purported “AGI” this simple IQ test question:

What is the shortest python program you can come up with that outputs:

0000000001000100001100100001010011000111010000100101010010110110001101011100111110000100011001010011101001010110110101111100011001110101101111100111011111011111

For background on this kind of question see Shane Legg's (now ancient) lecture on measures of machine intelligence:

https://youtu.be/0ghzG14dT-w?t=890

It's amazing after all this time that people are _still_ trying to discover what Solomonoff proved over a half century ago.

jabowery··on Hutter Prize Entry: Saurabh Kumar's “Fast Cmix” Starts 30 Day Comment Period
If you think that $100/month incentivises "plenty of investment" compared to the tens of billions put into less principled "SOTA" measures, I suppose that says all that needs to be said about your credibility.
jabowery··on Hutter Prize Entry: Saurabh Kumar's “Fast Cmix” Starts 30 Day Comment Period
The failure is that of the machine learning world to underwrite the prize.

There is no greater potential risk adjusted return to the field.

$100/month paid out over the lifetime of the prize must be compared against the investments in machine learning over that time.

jabowery··on Hutter Prize Entry: Saurabh Kumar's “Fast Cmix” Starts 30 Day Comment Period
Saurabh Kumar has submitted "Fast Cmix" to The Hutter Prize for Lossless Compression of Human Knowledge.

The Judging Committee appreciates Saurabh Kumar's compliance not only with the rules of the contest but with the provision of an executable archive which makes our job easier.

As per the Hutter Prize Award's section of The Rules:

The contribution is subject to public comments for a period of at least 30 days before the prize is awarded.

jabowery··on Shortest meta-circular description of a universal computational structure?
It would be interesting to see someone like Tromp explore binary FOL the way he's explored binary lambda calculus*. There are good reasons for going the FOL direction since functions are degenerate relations. While it is true that functional languages based on, for example, lambda or SK calculus, can naturally support "and parallelism" (eg x^2 in parallel with y^2 in x^2+y^2), getting "or parallelism" requires expression of independent processes (e.g.indeterminacy), some of which may not terminate. For example, operating systems need "or parallelism".

*I'm not at all comfortable with Tromp's abandoning SK for lambda calculus in his search for a principled choice of UTM for Algorithmic Information Theory. His justification seems to be that lambda is "more expressive" than sk, but that begs the question: To express _what_?

jabowery··on Shortest meta-circular description of a universal computational structure?
Yep. Tromp is indeed a legend I recall running across back when the Hutter Prize was first introduced and people were making noises about the "arbitrary choice of UTM" as though that was a good reason to not use lossless compression as the most principled loss function for machine learning. Indeed, he's still with us and gave this interview last month! https://youtu.be/ejhfJScuViY
jabowery··on Shortest meta-circular description of a universal computational structure?
To quell AGI hysteria I've posed a simple Algorithmic Information Theoretic IQ Test question: What is the shortest program you can come up with that outputs this string: 0000000001000100001100100001010011000111010000100101010010110110001101011100111110000100011001010011101001010110110101111100011001110101101111100111011111011111

Yann LeCun responded: "I can construct a computer for which the shortest program that generates this bit string has length 1."

To which I posed this question: "Why do mathematicians spend centuries asking for their minimum set of axioms but, not even Wolfram's acolytes ask for the shortest meta-circular description of a universal computational structure?"

jabowery··on [dead]
It may be unfair to the LLM community to pick on them over their ignorance of Algorithmic Information as model selection criterion since this model selection criterion has been inexcusably ignored by the entirety if the natural sciences for over half a century. However, the time is getting short and if any community should be introspecting about its ethics and integrity regarding bias at this point, it is the community that should be most familiar with information processing and algorithms.

Consider this essay "tough love" as this community is driving us all into the abyss.

jabowery··on Investors Serious About Application of Solomonoff Induction to Model Selection?
"Tendencies of physicians in training to be overconfident have to be taken into consideration to ensure the appropriate degree of supervision and feedback."

https://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect

I don't know that my diagnosis is correct, but it certainly seems to be the case that your continual referral back to AIXI despite it being excluded from consideration in my original post and then re-excluded in my most recent comment indicates that your reading comprehension (including my "speech patterns") isn't nearly what you think it is. Of course, another explanation is that you are simply a troll, but that is a decreasingly likely explanation.

jabowery··on Investors Serious About Application of Solomonoff Induction to Model Selection?
Thank you for your medical diagnosis of my speech patterns. But the fact that you conflate the 2 decade old AIXI with the 5 decade old Solomonoff Induction despite my attempt to filter out ignoramuses with the 2 fallacies (neither of which mention AIXI and one of which specifically singles out AI/AGI for elimination), I'm afraid you've just been filtered, along with your medical diagnosis.
jabowery··on Investors Serious About Application of Solomonoff Induction to Model Selection?
Yes, I know what your point is. I'm just wondering if the distribution of "VCs" has a fatter tail than you think. The very name of this website, "ycombinator" is rather "on the spectrum" but it is obviously less so now than it was, say, back in the early 80s when Haskell Curry signed my sigplan proceedings. Likewise, back in 2006 when Marcus sponsored the Hutter Prize, it was rather "autistic" to think there would be anyone who understood what he was talking about. Then his "autistic" PhD students founded DeepMind and everyone went nuts about AI. I could cite dozens of examples of the way distributions shift in directions that seemed at one time to be only for "autistic savants".
jabowery··on Investors Serious About Application of Solomonoff Induction to Model Selection?
The flip side of that wisdom is that investors don't invest in technologies, the markets for which they know nothing about. See "technical market research".
jabowery··on GPL Release of IVR Implementation of Liquid Democracy
In addition to an LD implementation, it provides a local currency system providing mutual support for participation.
jabowery··on Lossless compression to avert mass bloodshed?
Matt Mahoney, one of the leading lights in lossless compression, comments over at the (high noise level) AGI forum, and I respond with a scenario illustrating how this might reduce the likelihood of a near-term breakout of massive neighbor-against-neighbor bloodshed emerging in Virginia over gun control:

https://agi.topicbox.com/groups/agi/Tf67f6c4584fac2f7-M05a4c...

← PreviousPage 2 of 2