wufufufu

There are no clear details about what DOGE will alter. All the information we have so far amounts to outrage-bait one-liners. Where and how are they deciding to make cuts, and which agencies are they planning to change? Cutting is not intrinsically a good or bad idea, but everything I’ve read about this so far makes DOGE seem anti-intellectual.

> And then I tried gpt-3.5-turbo-instruct. This is a closed OpenAI model, so details are very murky.

How do you know it didn't just write a script that uses a chess engine and then execute the script? That IMO is the easiest explanation.

Also, I looked at the gpt-3.5-turbo-instruct example victory. One side played with 70% accuracy and the other was 77%. IMO that's not on par with 27XX ELO.

Kinda? Chess isn't solved. Complex problems can have better solutions discovered in the future.

Anyone have a technical writeup of the actual bug? I'm trying to explain how this could happen to people who think this is related to AI or cyber attacks.

What happened to the QA testing, staggered rollouts, feature flags, etc.? It's really this easy to cause a boot loop?

To me, BSOD indicates kernel level errors, which I assume Crowdstrike would be able to cause because it has root access due to being a security application. And because it's boot-looping, there's not a way to automatically push out updates?

Doesn’t make much sense for Hikaru to cheat. He competes over the board where it’s much harder to cheat, and he makes most of his money through streaming, not through competition results.

What about safety during accidents? It's better to be driving a heavier car in that case.

I would love if when buying an iPhone I was able to specify an entire class of apps that I could not install. For example, I could be banned at an OS level from a curated suite of social media apps. They could also make each app report the domains that it connects to so that the ban could continue at a DNS level. Apple is in a great place to do this because of their walled garden App Store model and withholding of root access from users. To undo it you would have to go into a store to use their proprietary software to change the type of apps you're allowed to install.

Don't we already have many answers? -- We just do nothing to address the problems.

Pragmatically, we need way more investment in environmental engineering and policies that put conservation first.

Since collectively we can't be bothered because we all need the newest electric vehicle, you can also accept that mass extinction is a byproduct of humanity.

Can someone explain why this is being downvoted?

> We document appearance effects in the economics profession. Using unique data on PhD graduates from ten of the top economics departments in the United States we test whether more attractive individuals are more likely to succeed. We find robust evidence that appearance has predictive power for job outcomes and research productivity. Attractive individuals are more likely to study at higher ranked PhD institutions and are more likely to be placed at higher-ranking academic institutions not only for their first job, but also for jobs as many as 15 years after their graduation, even when we control for the ranking of PhD institution and first job. Appearance also predicts the success of research output: while it does not predict the number of papers an individual writes, it predicts the number of citations for a given number of papers, again even when we control for the ranking of the PhD institution and first job. All these effects are robust, statistically significant, and substantial in magnitude.

The paper is paywalled, but that's the abstract.

I'm curious about how much of that is self-sabotage. I don't think we can rule out self-perception of attractiveness as a confounding variable. Internalized attractiveness? If anyone has a good study about that, I'd be very interested.

I'm at about $700/mth but I don't see why I can't make it to $2k monthly. It's a Slackbot [1,2] that assists user story refinement for Jira. If you're a SWE that uses agile methodology this may be interesting to you.

[1] https://slack.com/apps/A01EL1SR7V4-groomba [2] https://groomba.ai

Thank you for commenting these in. I find this quote "That when allowed to be themselves, their psychological issues largely disappear." interesting and hopeful -- and the way it's phrased suggests some causation.

Is there a study that proves this? I did a brief literature scan and it seems to me that researchers are unclear about the causation direction of things like trauma, psychological disorders, and GID/gender dissatisfaction. I think this is a pretty important part of the ongoing debate.

I don't want to be the "citations needed" guy, but I do have a tremendous respect for science and I don't want to just have to go with my gut on something like this.

Oh so there is pre-processing to find the useful portions? What are you using for the pre-processing?

I feel that it's inevitable that OpenAI et al. will be able to handle large PDF documents eventually. But until then I'm sure there's a lot of value of in this kind of pre-processing/chunking.

What OpenAI API calls allow sending these small chunks?

When you query something like "What is this research about?" is it able to use data from all chunks?

Lost me on "9 year olds"

I'd like legislation that acknowledges both the utility and harm that social media and more generally smartphones have. All I want is an internet-based texting only mobile-phone that doesn't have a browser or anything besides utilities. I don't want any feeds, reels, TikToks, etc. The fact that you basically need a smartphone to socialize, make payments, park, read menus, but it comes with a hamster-wheel for dopamine dispensing is like if the only way you could hydrate yourself is fountain soda -- you're just playing a game where you try not to get diabetes.

Given that the topic is pretty politicized and they didn't go into that much detail about controlling for possible confounding variables, I think the author is trying to express their personal beliefs through a study. I think the data is interesting but the title jumps to a conclusion that IMO the analysis doesn't support (yet).

> Across many applications areas, we’ll be left with a choice of using a 90% accurate model we understand, or 99% accurate model we don’t.

And how do you show that it is 99% accurate besides creating enough automated tests to the point that you could write the procedural version?

I think what I was missing from this article is how to evaluate a domain where neural nets or LLMs can be applied. Image-from-text generation is a great one because accuracy isn't strictly defined. However, telling ChatGPT "code this pacemaker for me" would have a real accuracy attached to it that you could confirm with unit tests.

Had me in the first half, lost me in the second. Why not try to create unions instead of "don't document anything because you're making yourself replaceable"?

Since the code is working and the main problems are the lack of documentation & structure, would it be possible to refactor & add comments with AI techniques? Or transpose it to javascript so we can play online?

You are missing something but not what you think

It sounds like the author enjoys programming in a vacuum, math, and theory. Those are basically useless for a software engineer. Being a subject expert and having the codebase and systems memorized are infinitely more valuable to a company than having general interest in programming.

I am enjoying this. Is there a way to pause? I have to verify each cell just to boggle my mind at the fact that everything is adhering to GOL rules.

What if we let people design their own content filters? That way the platform itself doesn't have to. You can easily select popular filters from a marketplace and the number one option will probably be "hide hate speech" or "hide all politics".

A significant amount of TikToks are simply video proof of harassment/bullying but I guess it's not a priority to them to detect and remove those.

Getting rid of my phone entirely. Although now I cannot do this because I have on-call for work. Does anyone have recommendations for a phone that is as well-manufactured and reliable as an iPhone but without data capabilities?

How is school going for you?

We are two software engineers frustrated with "grooming meetings" that were one-sided and ceremonial. So, we hacked together a polling engine for asynchronous voting, satisfying the needs of our day jobs. Since then, we've seen companies adopt Groomba in different ways, and after implementing some feature requests, we realized we created something highly adaptable to many workflows.

We're offering Groomba free for your first team of users (however many people you want in a Slack channel) as we're confident Groomba can easily fit into your agile process. If not, just message us :)

The shrooms community has known about this for a long time, and now it's just about recreating the knowledge in peer-reviewed studies. Although there is a lot of "woo woo" on shrooms usage in the online communities, I do hope that researchers can extract the useful bits.

This isn't like an SSRI. 25mg psilocybin isn't something you can just pop at the beginning of a workday. Your setting and "head space" matter a lot for how your "trip" (which we will now rebrand as "treatment") will go. This needs to be accounted for in future studies.