Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
lucasfin000
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
5 ms
·
1.
▲
by
lucasfin000
5mo ago
This is definitely a fair concern, and something that we have thought thoroughly about, but let me clarify some things: Our architecture makes jurisdiction less relevant than it would be for a traditional email provider. All email content,
2.
▲
by
lucasfin000
5mo ago
Yes, we did see the thread shortly after it was posted, and we did move the restricted email address validation to the server side. The client-side check is still there in the UX layer, but it is no longer the security boundary. Thank you f
3.
▲
by
lucasfin000
5mo ago
Hi HN. We have been building a quantum-safe, end-to-end encrypted email service where we, by design, cannot read your mail. Very few encrypted email services have post-quantum cryptography in production that works with any encrypted email p
4.
▲
Aster Mail – End-to-end encrypted email with post-quantum cryptography
(astermail.org)
1 points
by
lucasfin000
5mo ago
|
7 comments
5.
▲
by
lucasfin000
6mo ago
The "just buy another one" argument only works if the alternatives are even comparable. For a lot of people, macOS is a hard requirement and not a preference, so telling them just to buy a framework that runs Linux ignores that en
6.
▲
by
lucasfin000
6mo ago
The real frustrating part is that Cloudflare's "definition" of suspicious keeps changing and expanding. VPN users, privacy-first browsers, uncommon IP ranges, they all get flagged. The people most likely to get caught by thes
7.
▲
by
lucasfin000
6mo ago
This is definitely the right question. A list of failures without any baseline won't tell you anything. You would need the same exercise for human-written code at a comparable scale before drawing any conclusions at all. Without it, it
8.
▲
by
lucasfin000
6mo ago
The actual scariest part isn't that the AI got it wrong... it's that nobody felt the need to verify the AI. A tip from an anonymous caller can get investigated and found out if its true or not, and a match from a facial recognitio
9.
▲
by
lucasfin000
6mo ago
Post-quantum crypto is a good example of this. Lattice-based schemes were theorized in the 90s, but they took decades to actually reach production. The math existed, the hardware existed, and the ideas for making it work were just not there
10.
▲
by
lucasfin000
6mo ago
MoE feels a lot more like engineering to me. You're routing around the problem rather than actually solving it. The real math gains are things like quantization schemes that change how information is actually represented. Whether that
11.
▲
by
lucasfin000
6mo ago
The asymmetry is what makes this very interesting. The cost to inject poison is basically zero for the site owner, but the cost to detect and filter it at scale is significant for the scraper. That math gets a lot worse for them as more sit
12.
▲
by
lucasfin000
6mo ago
The tone and sensitivity thing is a real issue. A neutral prompt will get a neutral answer, but adding any emotional charge, it will immediately fold. That's not really a reasoning failure it's just a training problem. RLHF reward
13.
▲
by
lucasfin000
6mo ago
uBlock origin on Firefox or Brave, which will block most of the tracker bloat, causing the RAM spike. It's not a perfect fix, but it will cut out a significant chunk of it. Tab Wrangler also helps by suspending inactive tabs automatica
14.
▲
by
lucasfin000
7mo ago
This is the price every small sender pays. The unblock request process is essentially designed to make you give up or move to a large ESP. There's no appeals process, no SLA, no acknowledgment that your reputation data might just be wr
15.
▲
by
lucasfin000
7mo ago
The awkwardness after your answered was the interview telling you something important. A team that penalizes picking the right tool over an impressive one is a team where you'll spend years creating complexity nobody needs. The lesson
16.
▲
by
lucasfin000
7mo ago
I dont think the argument is really about child safety. If it was tiktok would also be working on fixing their algorithm that can send minors toward harmful content, which is a far larger documented vector than encrypted DMs. This is about
17.
▲
by
lucasfin000
7mo ago
That's the entire point of verified boot with custom keys, you don't need to trust Motorola or Lenovo. You can control what runs from the first boot, the threat model for a compromised supply chain is different from a backdoored c
18.
▲
by
lucasfin000
7mo ago
The worst part isn't even that quote, its that nothing structurally has changed one bit since then. The business model still requires users as the product. Glasses that upload video to Meta's servers is the entire point.
19.
▲
by
lucasfin000
7mo ago
Even if the algorithm holds up we are still years out from a actual quantum computer that can run at scale. But that's kind of the point. NIST finalized ML-KEM in 2024 because you don't need to wait until the house is on fire to b