veunes

17 ms·

151.
by veunes 8mo ago
The zeroize after exec feature sounds good, but what is the threat model in an agent context? If the agent can run printenv in the first millisecond and exfiltrate it (if net is allowed), zeroizing won't help It seems egress filtering
152.
by veunes 9mo ago
Not simple in the sense of easy, but simple in the sense of foundational: if a government can't even roughly say how many people it governs, everything built on top of that gets shaky
153.
by veunes 9mo ago
I think you're right in principle, but the article is pointing at a slightly different failure mode than just "wide error bars"
154.
by veunes 9mo ago
Yet a lot of the global dashboards we argue over are built on numbers that might be wrong by tens of millions, and we rarely stop to ask how those numbers were actually produced
155.
by veunes 9mo ago
Running the code checks if it works now, whereas code review checks if it will work in a year and if anyone else can understand it. Tests don't catch architectural mistakes or time bombs. If you remove reviews and rely solely on tests,
156.
by veunes 9mo ago
The main problem with current AI reviewers isn't catching bugs, it's shutting up when there is no bug. Humans have an intuitive filter like "this code is weird, but it works and won't break prod, so I'll let it slid
157.
by veunes 9mo ago
One doesn't exclude the other. To fix something you first have to validate it and figure out exactly what the AI got wrong. But yeah, you're basically right
158.
by veunes 9mo ago
That's fair, AI handles spec-compliant code quite well. The sanctuary for humans remains where the map doesn't match the territory: debugging silicon errata, hardware-level race conditions, or working with custom, proprietary gear
159.
by veunes 9mo ago
OnePlus is a textbook case of that quote
160.
by veunes 9mo ago
When a remote update can irreversibly change hardware state, ownership becomes conditional
161.
by veunes 9mo ago
What's worrying isn't the CRA itself, but that companies may use it as cover to lock things down more than necessary
162.
by veunes 9mo ago
On most devices, anti-rollback means "older firmware won't boot" or "you lose secure features." Here it seems to mean "try it and you permanently brick the device," with no warning in the updater and no pu
163.
by veunes 9mo ago
If this becomes the norm, it effectively ends the idea that you own the hardware you paid for
164.
by veunes 9mo ago
That's a dangerous distinction in the AI era. If you reduce your work to solving problems given a set of requirements, you put yourself in direct competition with agents. LLMs are perfect for taking a clear spec and outputting code. A
165.
by veunes 9mo ago
The only way to be bulletproof is to be the person who takes ownership. AI can generate an app, but it can't answer to a court, clients, or the CTO when the database crashes on Black Friday. Shift from writing code to risk management.
166.
by veunes 9mo ago
Fair, but the threat model matters here. For a static mortgage calculator, the data leak risk is zero (if it's client-side). The risk here is different - logical. If the AI botches the formula and someone makes a financial decision bas
167.
by veunes 9mo ago
I feel you. There's a massive difference between crafting and assembling. AI turns us from artisans carving a detail into assembly line operators. If your joy came from solving algorithmic puzzles and optimizing loops, then yes, AI kil
168.
by veunes 9mo ago
The key phrase here is "I still had domain expertise". Many miss that AI is a multiplier. If you multiply 0 by AI, you get 0 (or hallucinated garbage). You multiplied your knowledge of compound interest and UX by AI's speed.
169.
by veunes 9mo ago
There's a health and capacity angle. A lot of today's grandparents are still working, dealing with their own medical issues, or simply don't have the energy to provide full-time childcare
170.
by veunes 9mo ago
When people say "there are barely any kids," they're often describing the outcome of past policy choices, not a reason to avoid changing them
171.
by veunes 9mo ago
Subsidizing childcare helps families stay, but it doesn't address why childcare, housing, and everything else are so expensive in the first place
172.
by veunes 9mo ago
Reading works when you generate 50 lines a day. When AI generates 5,000 lines of refactoring in 30 seconds, linear reading becomes a bottleneck. Human attention doesn't scale like GPUs. Trying to "just read" machine-generated
173.
by veunes 9mo ago
The tools exist, they're just rarely used in web dev. Look into ApiDiff or tools using Tree-sitter to compare function signatures. In the Rust/Go ecosystem, there are tools that scream in CI if the public contract changes. We need
174.
by veunes 9mo ago
I totally get the fear regarding probabilistic changes being reviewed by probabilistic tools. It's a trap. If we trust AI to write the code and then another AI to review it, we end up with perfectly functioning software that does preci
175.
by veunes 9mo ago
For a lot of people postponement isn't a strategic move
176.
by veunes 9mo ago
It's hard to imagine a parenting strategy with no tradeoffs
177.
by veunes 9mo ago
The tricky part (both as a parent and a manager) is knowing which fires are safe to burn and which ones will burn the house down
178.
by veunes 9mo ago
From the outside, "strategic delay" and "indecision" can look identical, but they feel very different to the people depending on you
179.
by veunes 9mo ago
The technique is very role-dependent. It works well when (a) others are capable of resolving things on their own, and (b) the cost of delay is low and bounded. In environments with unclear ownership or asymmetric information, delaying can j
180.
by veunes 9mo ago
If it were just an average, we'd only get gray sludge. Models learn the manifold distribution - they don't just mix existing works; they discover the hidden rules by which those works were created. This is reverse-engineering of h

More ›