yesb

>RAG on the other hand creates the hardest challenge for these models. It instead puts 5 ideas with the highest similarity into the context in full.

I think what you're observing is that there is more to information retrieval i.e. "retrieval" in RAG than slapping everything into a vector database and calling it a day. There's no such requirement in RAG to mindlessly load the k nearest neighbors into your context and see what happens. That's a very rudimentary implementation.

This markdown system I'd argue is RAG as well. You're just doing the retrieval in a way customized for the problem at hand. If you have a precise method of retrieving the most relevant things, obviously use that rather than a similarity metric. If I'm reading correctly, this markdown system is basically a knowledge graph which is not a new idea.

>Unpopular opinion around these parts, but there's a high correlation between the best candidates and people who take the time to review the job posting and customize their resume before submitting.

There's also a high correlation between matching the job description perfectly and the resume being AI generated and/or dishonest. Doubly so if the requirements are overly specific or contain pure nonsense as some do (saw "Must have 5 years experience with Claude Code" recently). You can filter out people who blatantly don't match the job and those who obviously just copied the job description. Never seen much correlation between the quality of resumes in the middle (ok to great) and how well they do in interviews.

Not to mention at many places "requirements" are mere suggestions or a low-effort list of technologies taken from the resume of the person who quit. Everybody has their own filtering rules and hiring managers rarely agree with each other much less the HR people usually doing the first round of filtering. The true requirements come out only at the resume filtering or interview stage.

>I doubt the "best candidate" is also the person who can't be bothered to take 10 minutes to update their resume before pressing submit.

The "best candidate" a lot of times doesn't mean the best able to do the job. It often means the ones who 1. makes the hiring manager feel good by being enthusiastic about their (statistically) average job for average pay and 2. minimizes their perceived risk of looking bad for their hiring choice.

>If a job listing asks for Python and someone submits a resume that does not list Python, it's fair to reject that candidate.

If you're not a FAANGMULA type place then sure. But also if your job posting lists something like "Python 3.14" or "7+ years Python" I know the hiring manager probably doesn't know the first thing about python. In that case I think it's fair for any decent programmer to lie because they will likely surpass the hiring managers knowledge by studying python a bit before the interview.

No but hole punching isn't really needed in that scenario. Open a port on the IPv4 side + dyanmic DNS. More than likely that is what tailscale is doing for you. Opening a port on that end w/UPnP. But can't argue with how simple it is with tailscale.

Employment green cards are sometimes obtained via consular processing too.

Doubtful. Most consular green card applications are for family based immigration. This pause is also for final interviews for visas that are basically approved already. Labor market test already happened a while ago if relevant.

Employment based green cards are mostly granted inside the U.S. via adjustment of status without state dept involvement.

SSN card + ID works for employment purposes.

Simply applying for renewal or naturalization extends the card as well.

>successive administrations ignored the intent of the law, for so long that people began to rely on assurances that the law never promised. Now that the law is getting applied properly, those people are facing the consequences.

No, there's nothing proper about it and it has nothing to do with the merits of the visas in question. There is no law saying that the state department should make travel impractical for people USCIS has already decided can live/work in the US. The reverse is much more likely. They want to push people off of legally obtained visas by causing unreasonable delays.

H1B is temporary but you can have immigrant intent. It can be extended indefinitely with an approved green card application. Your employer is going to pursue that if you're on h1b and they want to keep you around. Many if not most h1b holders are in that category and will get a green card eventually. They are simply waiting for that to be available due to quotas based on their qualifications and birthplace.

Not sure what you're trying to say regarding back-dating. I think what you're missing is that long term holders of H1Bs universally have an approved PR or at least a pending one. That's the basis for extending it past the usual temporary limit. They need to stay on the work visa (with its limitations) until the immigrant visa becomes available (different from approval).

>seeking permanent residence on a tourist visa is not allowed, under most conditions.

In the US, it's allowed under your stated conditions. Immediate relatives of US citizens can generally adjust status to PR from any status (including tourist visa/visa waiver) or no status. Although if your underlying status expires you could be temporarily detained on a discretionary basis.

That said, it's better to pursue a K1 if you plan to get married shortly after arriving. It's very unlikely you will be granted a tourist visa if you tell the truth in the interview that you're engaged to someone in the US with immediate plans to marry.

Border searches aren't arbitrary when done for the purposes of border security and applied to travelers in a non-arbitrary way. It's pretty clear that CBP agents have a wide discretion to search people at their own whims i.e. arbitrarily.

>that the government does so lawfully

There are many legal question in this case, it's not clear the government acted lawfully. If he'd have arrived in a different federal court district, the search was unlawful because precedent says they need reasonable suspicion.

Not to mention this was a pretext to gather warrantless evidence for the FBI for a domestic investigation and nothing to do with securing the border.

>Antagonising law enforcement in this situation is not a smart move

Not what he did.

Generally I think the person implementing it has a poor native scrolling experience due to poor hardware and/or software and the hijacked behavior seems better to them.

The "narrative" above is that they're both made by seemingly well-intentioned solo developers, that's it. Use your own judgement for something that wants to inject JS into every website you visit.

I have been happy with my workaround for this issue which is using uBOL along side the AdGuard content blocker extensions (main AdGuard web extension disabled).

A number of games (for example the new Halo) aren't actually on the disc or otherwise have mandatory download requirements. The disc is just used as a DRM mechanism. Rockstar is not shipping discs for completely non-technical reasons. Likely some combination of saving a few bucks, stopping people camping out at retailers, and blocking the secondary market.

It does but it's not reliable. The update process is also the slowest imaginable other than Windows update. It takes several minutes at times just to discover which games need updates.

Had the same thought but with the new Halo remake which I own on PS5.

On booting the game, you're presented with the text "Press X to join" in the second player slot. That's a lie, pressing X just makes the screen flash.

Had to search online what to do. Turns out you need a Playstation account AND Microsoft account, AND they need to be linked just to join as a split-screen player. However, no prompt or UI is implemented to explain or do any of those for a second player.

Ok, setting out to do this BS. Feeling nostalgic about Halo CE LAN parties with 0 internet connection or accounts needed. She logs in with the PS account which she had to make to play another couch co-op one time a year ago. That itself was for no reason other than to boost Sony's MAU metrics but I digress. MS account made. I assume you need to create an Xbox gamer tag too, so we do that. Back to Halo, same flashing screen issue. No prompt appears to link the PS account to the MS account. I have a hunch that MS is incompetent and didn't implement the linking process except for the primary player. I'm right. You have to force close the game, switch the primary PlayStation profile to the second player, then relaunch the game. Finally got the prompt to link accounts. Now "Press X to join" works. Truly egregious how poorly implemented and communicated this is.

At this point we're 45 min in and have lost interest.

>I don't want to know what life is like having xbox hardware

For several years, you've been forced to login with a MS account on first boot.

Flock's system misread the plate, simple as that. Trying to pitch it as solely human error just shows that they don't want to take responsibility of the limitations of their software. Handling partial matches should be the bare minimum. In this article the CEO talks like considering those is a great idea that they'll look into. That's so incredibly dumb that I can't believe that they don't already do that. What likely actually happened is that the plate was incorrectly flagged as a perfect match.

Why is their tech not tested on every plate format that is currently used? That should be table stakes. Why did it not flag the match for review because the "34" and "DTM" had an unusual amount of space between them (not matching expected spacing of NJ plates)? The truth is that they are probably just using some free ML models off of hugging face and selling it as a magic black box that finds criminals to police.

They are listing what the position entails. In this case it's a full-stack web dev role. "only c++ all their life" is more or less not possible if you have experience doing that

Likewise from another job post: "Have a strong background in kernel-level systems". Not possible if your whole resume is building web apps, may be possible for someone that has only used c++ professionally.

HDMI ethernet channel is a thing, though semi-obsolete. It's unusual for a PC graphics card to support it. Intel website suggests they don't support it: https://www.intel.com/content/www/us/en/support/articles/000...

Why is not connecting your TV to wifi asinine? Generally works fine but I suppose there are rumors that some TVs scan for open networks and connect to them automatically.

> The only solution I suggest is physically removing WiFi cards from the guts before turning on.

It's going to be very unusual to find a TV using removable PC components like wifi cards. Another option is to connect it to your network but block it from the internet

Tried out the wg-wrap tool, might come in handy from time to time. Neat that it was made with a local model.

Some issues:

1. `wg-wrap healthcheck` was all green even though unprivileged user namespaces was restricted via AppArmor (Debian). That check doesn't seem to work

2. DNS doesn't work (no domains resolve) if the config lists multiple servers e.g. DNS = 1.1.1.1, 1.0.0.1

3. Peer endpoints don't support domain names, only IP addresses

4. Minor: the tool doesn't add an implied /32 cidr prefix for single ip configs (common from some VPN providers).

I would not be surprised if the main purpose of these job openings is to boost sign-ups/download stats. It's "weird" because generally it's considered poor taste to extract commercial value from people applying for a job in good faith. Especially before any kind of mutual interest or human interaction has been established.

Elevated creatinine levels (in absence of supplementation) is a sign of kidney issues. There are many studies showing no reverse link i.e. supplementing creatine damaging kidneys

https://link.springer.com/content/pdf/10.1186/s12882-025-045...

That may exist but the opposite type of irrationality is much more common.

Scalability = success. We need to be "scalable" because that means we're successful right? Scalability = real engineering. I'm a real engineer so I need to design everything to be "scalable" because I'm so smart

>The funny part is when one of those people fluke themselves into senior management when their saas takes off.

>Inevitably they have to suck it up and hire experts in the same technologies that "no one needs".

Sounds like they were the wise ones to build something simple that achieved a high level of success.

I remember vaguely the opposite. They were the only major prediction site that thought Trump had any significant chance of winning.

This article is relevant to a theoretic perfectly designed and built guitar.

Relating it to performance is just silly. Most companies barely understand the performance of their employees much less candidates. The market has shrunk but not catastrophically so. Most people haven't been majorly affected but that doesn't mean they're automatically the most deserving or best performing.

People with experience and/or credentials desired by companies in areas of growth (i.e. AI) are always in high demand

Sure, I assumed status quo everyone is talking about is basically the several years before that graph. I still think it's relatively bad compared to that despite the modest improvement.

What's not shown in a graph of job postings is the demand side. With all the layoffs, out of work college grads, people staying put in jobs they are unhappy with, etc., I'd wager that demand per job is still at a historically high level compared to what we have been accustomed to

You'd think "past year" would include a full 12 months. This person has chosen a ~10 month period to hide the large drop off in early 2025 as you can see here: https://www.citadelsecurities.com/news-and-insights/2026-glo...

Their graph shows a rebound to early-mid 2024 levels which is promising but still a relatively bad job market

If anyone is curious, as I was, where this misinformation came from: it appears to be a criticism of the Food Compass rating system from Tufts University. The connection to "past administrations" seems to be added by the person I'm replying to. They've also swapped Cheerios with Cheetos.

>On social media, I have seen graphics showing certain breakfast cereals scoring higher than eggs, cheese, or meat. Did Tufts create these graphics?

>No. Food Compass works very well, on average, across thousands of food and beverage products. But, when this number and diversity of products are scored, there are always some exceptions. These graphs were created by others to show these exceptions, rather than to show the overall performance of Food Compass and the many other foods for which Food Compass works well. But, as objective scientists, we accept constructive criticism and are using this to further improve Food Compass. We are working on an updated version now – see our versions page for more information.

https://sites.tufts.edu/foodcompass/ https://www.nature.com/articles/s43016-021-00381-y.epdf

> The good engineers in india know their value and get it.

The job market is not that efficient anywhere, especially India. Lots of people work their way up from crappy jobs to good ones, just like in the US.

>My company has offices in india because you have to manage them yourself not use middlemen. You can train the locals to be great managers (at least some).

"some members of this primitive tribe can be taught our sophisticated ways"

The issue with middlemen is they are basically labor arbitrageurs and have an incentive to hire the cheapest people possible and inflate their credentials/abilities. Same thing happens with onshore consulting firms.

>wages for good people in india are worse similar people in the us, but often high than in europe.

"Often higher than Europe" is a stretch. Typical big co with an India office pays maybe 20-30k USD per year for an engineer. And that is a good job relatively speaking. Top tech companies pay more but they also pay more in Europe