lanternfish

5 ms·

1.
by lanternfish 1mo ago
For a given short-context (hash function seeding) you do have detectable manipulation of the completion (how they can read the marking). But, because (assuming enough entropy in context) the hashing function is decoupled from the log-probs,
2.
by lanternfish 1mo ago
There are no watermark related tokens, there is no watermarked set - read the paper, it's public and not that complicated. It doesn't change the distribution of completions, and won't change the situation where one output tok
3.
by lanternfish 2mo ago
If it really was the case that the harness is at fault here, that'd be much more terrifying, as it would imply that current frontier-adjacent open models (like K3) would be able to achieve similar ends with a custom maladapted harness.
4.
by lanternfish 5mo ago
10 Axios's within 5 days.
5.
by lanternfish 6mo ago
Why not?
6.
by lanternfish 6mo ago
Engineering is done in the context of constraints, cost is one constraint - and its a relatively conserved constraint. Saving labor in one area allows for more care in other areas. Especially given that labor is often not cost constrained,
7.
by lanternfish 7mo ago
Mobile app compatibility is presumably the biggest seller.
8.
by lanternfish 9mo ago
In Iraq, JSOC operational doctrine was literally to target assassination campaigns based on 'nodal analysis' from contacts lifted from cellphones; if telegram had existed, they certainly would have used it too. Famously, they didn
9.
by lanternfish 9mo ago
That wasn't my point. My specific argument is about US operational policy on the ground in similar engagements. Based on precident, we would expect them to engage in the behavior the commentor indicts. I dream that neither of these imp
10.
by lanternfish 9mo ago
See: Panama, Vietnam, Cambodia, Afghanistan, Iraq, etc.
11.
by lanternfish 1y ago
The problem is the same problem with crypto dao projects - cryptographic certainties only apply to mathematical structures; you can't validate that someone actually holds a quality until you can embed that digitally. That turns out to
12.
by lanternfish 1y ago
In a mathematical sense - absolutely. You can dual halting problem against many very tangible qualities - like whether a (proved) statement is true or false. A (large-n) halting program is closer to an instantly halting program not just bec
13.
by lanternfish 1y ago
An LLM - which has functionally infinite unverifiable attack surface - directly wired into a payment system with high authentication. How could anyone anticipate this going wrong? I feel like everyone is saying 'we're still discov
14.
by lanternfish 1y ago
I think that's called Myst (1993)
15.
by lanternfish 2y ago
`M(ultimodal) Ag(ent) [ma]` maybe
16.
by lanternfish 2y ago
I'm honestly not sure that their analysis passes muster. It seems that the main consideration is that Harris underperformed compared to down-ballot races and that the underperformance was ahistoric. However, the campaign was also ahist
17.
by lanternfish 2y ago
I agree that it's not the best term, but I don't think its so disqualifying that it makes the claim untrue: it's misleading at worst, and that imprecision only kinda interacts with the underlying claim. I guess the better phr
18.
by lanternfish 2y ago
"Recorded history" in the title refers to the period of history where the agency has been recording the numbers. It might not be the best phrasing, but it's not strictly untrue; the underwritten thesis (TB is on the rise) is
19.
by lanternfish 2y ago
The actual work underlying the essay - the one published in Cambridge Core - is pretty strong and has a lot of pretty compelling analysis. It's just long.
20.
by lanternfish 2y ago
This seems an insufficient analysis. The meaning expressed by contemporary music, film media, or streaming television isn't very profound, but they at least still make a passing effort to "signify" something. The highest gros
21.
by lanternfish 2y ago
Your conclusion is the exact same one the above essay comes to in the next paragraphs. It concludes that the "alignment" between MrBeast and his work is a result of the larger thesis: creators are ultimately created by the audienc
22.
by lanternfish 2y ago
Contact Mike is one of the most important encyclopedia checks in the game, honestly. Its one of the few direct hints you get to HDB's backstory before the literal last scene, which can actually give you actionable insight in a couple s
23.
by lanternfish 2y ago
Oh shit I totally misread the policy - I interpreted "payments to such accounts" to mean donations etc. made through channels that the host supported. As written, it's not really an offset, and really just a way to wash hands
24.
by lanternfish 2y ago
They hope to offset some sort of imperative burden (presumably moral) of hosting onerous content by countermanding the effect of hosting that content with paired monetary support of its adversary. My consideration is that pairing effect the
25.
by lanternfish 2y ago
Maybe its unfair pessimism, but I definitely believe that Kiwifarms (ex) is way more efficient at turning money into targeted hate than - say - the Trevor Project is at countering it. I guess my sense is that if you actually want to counter
26.
by lanternfish 2y ago
This relies on an EA adjacent market fallacy where we can resolve all moral action down to funding actors of various moral alignments - there's no reason to believe that the end utility (or whatever metric) of the action is linear w.r.
27.
by lanternfish 2y ago
you pay them in the only resource known to humanity to be worth more than money - petty forum power predicated on exclusivity.
28.
by lanternfish 2y ago
For anything weight bearing, I really wouldn't trust it for translation. More than once I've gotten an output that had subtly different meanings or implications than the input. A human translator is going to be significantly more
29.
by lanternfish 2y ago
I'd differentiate here between deductive and inductive reasoning. While english and history - through argumentation - expose some deductive reasoning, the vast majority of it is inductive.
30.
by lanternfish 2y ago
The problem is that mathematics education isn't just about learning times tables. It's also the primary medium schools use to communicate analytical reasoning and deductive analysis. If you cut math as a target without fundamental

More ›