Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
AlanYx
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
13 ms
·
181.
▲
by
AlanYx
2y ago
Does anyone have any recommendations for a good tool that allows both programmatic inspection and modification of PDF primitives. For example, let's say someone wants to iterate through every embedded image in a PDF and apply some form
182.
▲
by
AlanYx
2y ago
Psychometric intelligence ("g") is defined correlationally viz. cognitive tasks; that is, the definition is itself a correlation. But it's robust in the sense that it correlates beyond the correlations used in its definition,
183.
▲
by
AlanYx
2y ago
Yes, you could bring a suit if you knew about the interception. However, like FISA s.702, intelligence collection warrants under the French LI and German BND are generally secret, so most targets have no knowledge they are under surveillanc
184.
▲
by
AlanYx
2y ago
It's not that easy really. Several European countries have FISA s.702 functional equivalents that enable intelligence to get orders for interception of personal information on servers and entities within their legal jurisdiction. (e.g.
185.
▲
by
AlanYx
2y ago
I tend to agree with you about what will happen, but it illustrates the depth of legal uncertainty that exists in architecting software systems in Europe that process personal information. Corporations can't necessarily trust the EC&#x
186.
▲
by
AlanYx
2y ago
The EC has issued an "adequacy decision" regarding the new EU–US Data Privacy Framework (the replacement for Privacy Shield): https://commission.europa.eu/document/fa09cbad-dd7d-4684-ae6... and has begun &quo
187.
▲
by
AlanYx
2y ago
Thanks. That gives a lot more information. With respect to the new framework, the article you linked to just says "Since the end of last year, Uber uses the successor to the Privacy Shield." Do you know if the Dutch DPA endorsed t
188.
▲
by
AlanYx
2y ago
Thanks. So basically the new framework hasn't accomplished anything that can be relied upon when architecting a system to reduce the risk of GDPR compliance issues?
189.
▲
by
AlanYx
2y ago
Can anyone explain how this relates to the EU-US Data Privacy Framework (also sometimes called the Trans-Atlantic Data Privacy Framework)? I thought that that framework was supposed to allow this (as a replacement for the EU–US Privacy Shie
190.
▲
by
AlanYx
2y ago
Isn’t it a play on words/reference to Pudong in China? It’s a nice central part of Shanghai.
191.
▲
by
AlanYx
2y ago
Has Peter Thiel ever put his thoughts on higher education down in long-form writing? I've seen him speak about it, but I'd be interested in a deep dive.
192.
▲
by
AlanYx
2y ago
A good starting point is to try to build a pin-pallet escapement design, like Timex used in the 60s and 70s. This is largely doable at home with a 3D printer and basic tools, e.g.: https://incoherency.co.uk/blog/stories
193.
▲
by
AlanYx
2y ago
>Are you going to make the claim that there aren't 250 students who identified as Black, Hispanic, and/or Native American and Pacific Islander that can't get either a perfect score... Sometimes that is what happens with so
194.
▲
by
AlanYx
2y ago
Throughout the whole talk he's using an alternatively flippant/hyperbolic speaking style that probably has been effective for him in the past when addressing high-level non-technical policymakers. He speaks to their policy/st
195.
▲
by
AlanYx
2y ago
Are there any prominent examples of homomorphic encryption being used at scale in production? I hunted around for some a few years ago but couldn't find any.
196.
▲
by
AlanYx
2y ago
>I recently saw a literacy product aimed at parents, and which costs $50/month. They justify their pricing based on its substitutes being a tutor or private school tuition. There's a software startup aimed at teaching 3-5 year
197.
▲
by
AlanYx
2y ago
This isn't just folk wisdom either. There's an active subfield of cognitive science that focuses on working memory/cognitive load and math instruction. Broadly speaking, it's hard for students to successfully transition
198.
▲
by
AlanYx
2y ago
The most onerous parts of the AI Act, such as the provision that says "Member States shall confer on their market surveillance authorities the powers of... carrying out unannounced remote or on-site inspections", are not yet in fo
199.
▲
by
AlanYx
2y ago
It does say they have a business model ("our business model means safety, security, and progress are all insulated from short-term commercial pressures"). I imagine it's some kind of patron model that requires a long-term com
200.
▲
by
AlanYx
2y ago
He got new funding, but it's relatively modest ($2.69 million spread over two years).
201.
▲
by
AlanYx
2y ago
AFAIK there aren't any other major examples of directly elected governing bodies who don't have the power of legislative initiative. There are more examples if you count indirectly elected bodies like the German Bundesrat. The cor
202.
▲
by
AlanYx
2y ago
Depends how you define unchecked. The European Parliament's only real recourse against the Commission is censure, and it's never been successfully done (although it came close in 1999 (Santer) and in 2004 (Barroso)). On the other
203.
▲
by
AlanYx
2y ago
Thanks! If I'm reading that correctly, their best error rate with pre and post selection is 0.03%, but they end the paper with the statement "A significant milestone will be to demonstrate a universal family of quantum circuits wi
204.
▲
by
AlanYx
2y ago
The linked paper touches on quantum error correction but doesn't explain what the state of the art is. Has any team successfully demonstrated a single usable logical error-corrected qubit yet? I saw this article two months ago https:&
205.
▲
by
AlanYx
2y ago
There's a small DIY community who take XO-1 displays and installs them in old ThinkPads. It may not be as challenging as it first appears.
206.
▲
by
AlanYx
2y ago
Isn't it the case that you can only prove that a program halts if it can be reduced/transformed to a computational model (e.g., a model that is effectively calculable) that is less powerful than a Turing machine? Had they framed t
207.
▲
by
AlanYx
2y ago
The thing I've never understood is how Leike's superalignment scheme could work. There's a high-level description of their research aims here: https://openai.com/index/introducing-superalignment/ Ba
208.
▲
by
AlanYx
2y ago
It's got to be more than just a typical speedbooster, given that they have to adapt a comparatively huge focal plane to a lens aperture about 3mm in diameter. The article speculates that they use some sort of lens relay system, perhaps
209.
▲
by
AlanYx
2y ago
That's really an underrated part of the announcement. It sounds like they're moving the entire display TCON on-chip.
210.
▲
by
AlanYx
2y ago
M series have always been extremely competitive on single-core, but not really "the champ". Now this one is beating the Core i9-14900KS in a device with only modest copper heatsinking (with a heatsink likely less than 3mm thick)!
More ›