roughly

16 ms·

331.
by roughly 8mo ago
The moment you encode your biases in policy, you create vulnerabilities. What I’m hearing is that if I want to get something past your security policy, I need to route it through the Netherlands, possibly via a travel agency.
332.
by roughly 9mo ago
I think moving algorithmic feeds out of section 230 would do the trick. Once you’re curating the feed, you’re no longer a neutral platform.
333.
by roughly 9mo ago
Be interesting to see this compared with, eg, BMWs or something - see if this is a “rising tide lifting all boats” or a distributional signal.
334.
by roughly 9mo ago
I think there’s a good possibility that the technology that is LLMs could be usefully trained to decode binaries as a sort of squint-and-you-can-see-it translation problem, but I can’t imagine, eg, pre-trained GPT being particularly good at
335.
by roughly 9mo ago
Yeah, I don’t actually think LLMs are conscious, but this feels like the spontaneous firing of brain organelles that researchers experiment with - I don’t think those are conscious either, but it’s definitely getting outside of my “what if
336.
by roughly 9mo ago
Believe it or not, no - modern phones are rarely genuinely “off.” iOS has some form of “find my” that still works, I’m not sure where android’s at with it, but “off” and “not transmitting” are genuinely not the same thing at this point.
337.
by roughly 9mo ago
China was famous for low-quality products as well.
338.
by roughly 9mo ago
There was an article that came through here a little while ago describing the process by which commercial property owners and banks collude to keep storefronts boarded up and vacant because otherwise they’d have to adjust the loan terms or
339.
by roughly 9mo ago
Because PurpleAir is not a weather service, they’re a sensor company.
340.
by roughly 9mo ago
I've been pretty skeptical of LLMs as the solution to AGI already, mostly just because the limits of what the models seem capable of doing seem to be lower than we were hoping (glibly, I think they're pretty good at replicating wh
341.
by roughly 9mo ago
You know, it's funny - there's a couple places where this kind of thing seems to have come up. There's research around crop nutrition levels that shows decreases in nutrient levels as yield per acre goes up, there's rese
342.
by roughly 9mo ago
I’m… deeply suspicious of Gemini’s ability to make that assessment. I do broadly agree that smaller, better tuned models are likely to be the future, if only because the economics of the large models seem somewhat suspect right now, and als
343.
by roughly 9mo ago
One thing I’m becoming curious about with these models are the token counts to achieve these results - things like “better reasoning” and “more tool usage” aren’t “model improvements” in what I think would be understood as the colloquial se
344.
by roughly 9mo ago
FTA: > With a third group of participants, they used a non-invasive technique called transcranial alternating current stimulation to speed up or slow down the frequency of a person's alpha waves. And sure enough, this seemed to corr
345.
by roughly 9mo ago
Loudly broadcasting electronic signals out of something you’re trying to keep hidden seems like a tactical error, but these cats aren’t the best trained, are they?
346.
by roughly 9mo ago
The Americans who fought against the Francoists in the Spanish civil war faced enormous scrutiny back home for what was later described as “being prematurely anti-fascist.” The state worries about people willing to take up arms to protect t
347.
by roughly 9mo ago
> it was masked because our redundant comms channel made everything look nominal. Hah, this has bitten me often enough I check for it in test suites now - ok, you’ve proven the system works and the backup works, have you proven the prima
348.
by roughly 9mo ago
It’s sort of interesting to look back at ~100 years of the automobile and, eg, the rise of new urbanism in this metaphor - there are undoubtedly benefits that have come from the automobile, and also the efforts to absolutely maximize where,
349.
by roughly 9mo ago
As they say, any health study that doesn’t control for economic status is just studying the effect of economic status.
350.
by roughly 9mo ago
Holy shit that’s great. I need to check a few more of his videos.
351.
by roughly 9mo ago
> What's the "salient feature" that's missing? As it sits, all of the members of your coconut economy are going to be dead of malnutrition or exposure in relatively short order, so maybe address that and then we can w
352.
by roughly 9mo ago
> A simplified model is needed otherwise rigorous analysis becomes impossible If your tools aren’t capable of rigorous analysis of a model that retains enough detail to capture the salient features of the thing they’re trying to model, t
353.
by roughly 9mo ago
Alien Clay is also fantastic. I don’t want to spoil anything, but I think it gives the best intuition I’ve seen for a scientific concept that can be difficult to really grok otherwise.
354.
by roughly 9mo ago
> Suppose you're on an island where the economy only produces coconuts. This is why nobody takes economists seriously. What you lose in simplifying down to this model is literally everything. The coconut economy has zero predictive
355.
by roughly 9mo ago
These are still very early days for RISC-V, but I’m always happy to see things progress in this space. No, this isn’t a viable desktop for the average consumer, but if it makes the architecture more accessible for the types of weirdos who t
356.
by roughly 9mo ago
Do you have an Indow? I’ve been eyeballing them for a while, but they’re pricy enough I haven’t been willing to call the shot yet. I do have a pretty big bluetti and a couple panels for it - I also grabbed an RV fridge for storing food. Nat
357.
by roughly 9mo ago
If this is spiking some nostalgia for anyone, there’s a bit of a cottage industry in modernized motherboards for old thinkpads - eg, https://www.tpart.net/about-x210ai/
358.
by roughly 9mo ago
Be sure to watch the video itself* - it’s really a great piece of work. The energy is frenetic and it’s got this beautiful balance of surrealism from the effects and groundedness from the human performances. * (Mute it if you don’t like the
359.
by roughly 9mo ago
I’ve been using Waterfox recently, which feels almost nostalgic in how much it’s just a plain goddamn browser. It’s really delightful. Helium seems to be trying to be the same thing for Chrome - it’s replaced Brave as my go-to for the sites
360.
by roughly 9mo ago
> Use of analytics tends to replace user trials/interviews entirely, trading away rich signals for weaker ones Yeah, this is huge. The 30-day A/B test is a scourge on the industry.

More ›