faul_sname
Fuck around once, find out once. Do it again, now it's science.
No bio...
User ID: 884
faul_sname
Fuck around once, find out once. Do it again, now it's science.
6mo ago
It's not even clear that LLMs are analogous to cars here. When you call something a coder, I expect it to be able to do the job of a coder, rather than being a tool that helps improve performence.
The original tweet Jim referenced said
o3 is approximately equivalent to the #175 best human in competitive programming on CodeForces. The median IOI gold medalist has a rating of 2469; o3 has 2727.
Jim summarized this as
Apparently this AI is ranked as the 175th best coder on Earth.
Which is perhaps a little sloppy in terms of wording, but seems to me to be referring to coding as a task rather than a profession. I've never seen "coder" used as the word for the profession of people whose job requires them to write code, while I have seen that term used derogatorily to refer to people who can only code but struggle with the non-coding parts of the job like communicating with other people.
That said, if you're interpeting "coder" as a synonym for "software developer" and I'm interpreting it as meaning "someone who can solve leetcode puzzles", that's probably the whole disconnect right there.
From what you're saying they'd be more like high-performance component that could improve a particular car, but won't be able to go anywhere on their own.
Yeah, that's a good analogy. Coding ability is a component of a functional software developer, an important one, but one that is not particularly useful in isolation.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
10mo ago
faul_sname
Fuck around once, find out once. Do it again, now it's science.
1yr ago
Searching "debug interface", I see three places:
The first is on page 10, in section 3.1(Definition of terms, symbols and abbreviations: Terms)
debug interface: physical interface used by the manufacturer to communicate with the device during development or to perform triage of issues with the device and that is not used as part of the consumer-facing functionality
EXAMPLE: Test points, UART, SWD, JTAG.
The second is on page 20, in section 5.6 (Cyber security provisions for consumer IoT: Minimize exposed attack surfaces)
Provision 5.6-4 Where a debug interface is physically accessible, it shall be disabled in software.
EXAMPLE 5: A UART serial interface is disabled through the bootloader software on the device. No logon prompt and no interactive menu is available due to this disabling
The third is on page 32, in Table B.1: Implementation of provisions for consumer IoT security, where, at the bottom of the table, there is a "conditions" section, and "13) a debug interface is physically accessible" is the 13th such condition:
Provision 5.6-4 M C (13)
For reference
M C the provision is a mandatory requirement and conditional
NOTE: Where the conditional notation is used, this is conditional on the text of the provision. The conditions are provided at the bottom of the table with references provided for the relevant provisions to help with clarity.
So, to my read, the provision is mandatory, conditional on the product having a debug interface at all.
"But maybe they just meant that debug interfaces can't unintentionally be left exposed, and it should be left to the company to decide whether the benefits of leaving a debug interface open are worthwhile", you might ask. But we have an example of what it looks like when ETSI wants to say "the company should not accidentally leave this open", and it looks like
Provision 5.6-3 Device hardware should not unnecessarily expose physical interfaces to attack.
Physical interfaces can be used by an attacker to compromise firmware or memory on a device. "Unnecessarily" refers to the manufacturer's assessment of the benefits of an open interface, used for user functionality or for debugging purposes.
Provision 5.6-4 has a conspicuous absence of the word "unnecessarily" or any mention of things like the manufacturer's assessment of the benefits of an open interface.
So coming back to
They're still completely free and clear to have any interfaces for debugging or anything else that are meant to be usable by the user.
Can you state where exactly in the document it states this, such that someone developing a product could point it out to the legal team at their company?
faul_sname
Fuck around once, find out once. Do it again, now it's science.
1yr ago
You were off by a year.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2yr ago
I don't think "if we save the lives of our citizens it might cause moderate financial distress to our country" is a good reason to consign those citizens to death. Obviously there exists an amount of financial distress where it becomes worthwhile. "plunge the country into poverty to save one life" is obviously a bad idea. I'd argue even the recent case of "devalue the currency by 20% to prevent the loss of an average of 2 weeks of life" went too far on that front.
But I think curing aging would be a massive boon even considering the second and third order effects. I think it would probably be worth it even if it led to a full Zimbabwe-level economic blowup, though I'm not entirely sure in that case.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2yr ago
Surrogates exist.
I don't happen to have any lean beef jerky on hand, but I do have some dried shredded squid, which has 0.5g of fat, 25g of sugar, and 16g of protein per serving (and also, according to the back of the packaging possibly some lead, mercury, and cadmium?!), and that burns quite vigorously. That's rather more sugar (and heavy metals) than I would expect my dried squid snacks to have.
I do expect that even lean jerky would burn pretty vigorously once it got going though. I'll actually run the experiment the next time I have some lean beef jerky (that is not, for some reason, full of sugar and heavy metals).
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2yr ago
Do you really get the runaround on those sorts of questions? Because in my experience, if you give social sciences types any opportunity to talk about factors that could affect metrics of success by race / gender / immigration status / whatever, they will happily talk your ear off for hours. They are unlikely to mention genetic factors (outside of epigenetics and "did you know about DNA methylation [...] response to stress"), but that will not stop them enthusiastically brainstorming hypotheses and what studies one might run to test those hypotheses for as long as you're willing to listen.
Group outcomes are made of individual outcomes. Particularly in cases like "making friends with the right people", those individual outcomes may be correlated.
Not that "early luck compounded into long-term differences" can't be mitigated through social policy, but doing so transparently and fairly and in a principled manner is hard for the same reason that "controlling for" stuff in studies is hard.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2yr ago
I mean stonetoss being stonetoss isn't exactly funnier - here's the most recent one where it's just a low-effort dunk, vs this one, which is a bit funnier (though still low effort and not that funny).
On reflection I'd endorse both "the left can't meme" and "the right can't meme". Though is also possible that it's "nobody can meme in a way that people who don't spend all their time immersed in the same culture find funny".
faul_sname
Fuck around once, find out once. Do it again, now it's science.
1mo ago
I would say "more attention than puberty blockers", because the number of affected kids is much higher. Something caused a massive uptick in either the experience of dysphoria, the reaction to dysphoria, or some combination thereof. I think "social contagion" is a thought-terminating non-explanation here. To reduce the rate of trans identification, I think it would be worth looking into what generally leads to discomfort with being embodied (as that seems to correlate extremely strongly, and also seems to be much more common than it used to be).
Of course, if you don't actually care about that and your main objection is to "point deer say horse", that is perfectly valid. But in that event I also don't take statements of concern about puberty blockers at face value, and will discount your policy suggestions in that area accordingly.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
1mo ago
Oh nice. That is relevant to my interests in a way that I will share on Tuesday.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
1mo ago
Not a dedicated one, no, just a simple little 10 line Python script
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2mo ago
Oh good point. An all-wheel drive vehicle is a four-wheel drive under easy driving conditions, and a one-wheel drive if one of the four wheels loses traction.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2mo ago
I know they will lie about this and claim a massive drop in traffic regardless of what actually happens.
Do you think the NJ port authority is falsifying the EZPass data they're sharing here?
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2mo ago
For verbal I think it might be "I used to play a lot of Scrabble which trivialized the anagram task". I'd like to say my scores on the other tasks seem accurate except that for things like mental rotation I definitely struggle in real-world contexts (e.g. getting all the vector math right in a ray tracer) and yet I (and everyone else here who posted a score on that section) got a perfect score.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2mo ago
The relevant sample size is "at least one of 278", not "at most one of 100,000". But honestly if it was "we fucked up on this one of 278, but we're making a good faith effort to fix our fuck up" I think that would be fine. It's the "we fucked up, we admit we fucked up, we totally could fix it, but we won't and you can't make us" that is getting people up in arms.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2mo ago
It's trivial to identify examples of freedom that are not better the more people are free to exercise them.
Are there some specific freedoms do you think that you currently have but wish you didn't have? Alternatively, are there some freedoms you have and exercise, but wish you lived in a society where you and everyone else could not exercise that freedom?
faul_sname
Fuck around once, find out once. Do it again, now it's science.
2mo ago
@Glassnoser are you referring to this section?
- Every employer shall respect the worker’s right to carry on his activities in French; therefore, the employer is required, in particular,
(1) to see that any offer of employment, transfer or promotion the employer publishes is in French;
(2) to see that any individual employment contract the employer enters into in writing is drawn up in French;
(3) to use French in written communications, even those after termination of the employment relationship, with all or part of the staff, a worker in particular or an association of workers representing all or part of the staff; and
(4) to see that the documents below that the employer makes available are drawn up in French and, if also available in another language, see that the French version is available on terms that are at least as favourable:
(a) employment application forms;
(b) documents relating to conditions of employment; and
(c) training documents produced for the staff.
Despite subparagraph 2 of the first paragraph, the parties to an individual employment contract that is a contract of adhesion may be bound only by its version in a language other than French if, after examining its French version, such is their express wish. In the other cases, an individual employment contract may be drawn up exclusively in a language other than French at the express wish of the parties.
Despite subparagraph 3 of the first paragraph, the employer may communicate in writing with a worker exclusively in a language other than French if the latter has so requested. 1977, c. 5, s. 41; 2022, c. 14, s. 29.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
3mo ago
Superintelligences are going to be wayyyyyyyyy better at driving down those error bars and finding acceptable settlements. [...] I still don't see how a superintelligence doesn't reduce the bargaining friction.
I hope you're right about that. I worry that a lot of the dynamics around retaliation and precommitment are anti-inductive, and as such the difficulty of determining where the bright lines actually are scales with the sophistication of the actors. This would happen because a hostile actor will go right up to the line of "most aggressive behavior that will not result in retaliation" but not cross said line, so it becomes advantageous to be a little unclear about where that line is, and that lack of clarity will be calibrated to your adversaries not to some absolute baseline. And this is the sense in which I don't see a reachable point where honesty and bargaining come to strictly dominate.
As a note I do expect that bargaining frictions will be reduced, but the existential question is whether they will be reduced by a factor large enough to compensate for the increased destructiveness of a conflict that escalates out of control. Signs look hopeful so far but our sample size is still small. Certainly not a large enough sample size that I would conclude
The omniscient AIs will be able to plan everything out so far, so perfectly, that they will simply know what the result will be. Not necessarily all draws, but they'll know the expected outcome of war. And they'll know the costs. And they'll have no bargaining frictions in terms of uncertainties.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
3mo ago
By my assessment, poorly. In the sense that we will all be poor if things continue this way.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
3mo ago
He also campaigned on making America great and prosperous.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
4mo ago
I don't know if the grant application is public but here's the NSF page on the award, which has more details including the abstract and resulting publications.
Resulting publications look like real science with plausible important implications for medicine, not ideologically captured garbage:
- Tonic extracellular glutamate and ischaemia: glutamate antiporter system x c regulates anoxic depolarization in hippocampus
- Naked Mole-Rats Demonstrate Profound Tolerance to Low Oxygen, High Carbon Dioxide, and Chemical Pain
- Phenylpropanoid-enriched broccoli seedling extract can reduce inflammatory markers and pain behavior
- Pharmacokinetics of Injectable Meloxicam and Buprenorphine in the Naked Mole-Rat (Heterocephalus glaber)
This is exactly the sort of foundational research I want my tax dollars funding - low immediate commercial value but potentially massive positive externalities.
faul_sname
Fuck around once, find out once. Do it again, now it's science.
4mo ago
Previously, for example, Harvard grants on average charged 69% above the cost of doing research for institutional overhead. (I think we can all imagine where that ends up). NIH just capped that tax at 15%. This will save $4 billion per year. That's $53 for every one of the 75 million Americans who paid federal income tax last year.
If you follow the incentives off a cliff (as happened with health insurance), that means that if they want to retain their $4B cut, that means the new cost of doing research needs to be such that 15% of it is $4B.
That sounds like a concrete prediction. Care to make it concrete enough to bet on?
More options
Context Copy link