pwinnski

The entire point of an academic paper is to add to the sum of human knowledge. How can an LLM trained on a subset of human knowledge possibly even begin to accurate evaluate such a paper?

I trust an LLM to review that the language used in the paper is grammatically correct, but not to evaluate new information for accuracy.

One imagines that they're trying. I work at not-a-frontier-lab and my company still ended up hiring someone who apparently wasn't aware that we track system access. He was turned over to the federal authorities after trying to exfiltrate a lot of user+system data to North Korean IPs.

Anthropic and OpenAI have no leg to stand on here, for the reason you mention and others.

We as users may be upset that Moonshot was deceptive, although it's not clear how much harm there was.

Anthropic wanting us all to be upset on their behalf? No, thank you.

If Moonshot sold Claude access at Kimi prices? Seems like a win for users, albeit a deceptive one.

Polls use sampling to avoid the sort of social bias suggested by your question.

Ask the right person, like someone who protested, and they could tell you that literally everyone supports the protests. Why, everyone they know participated in at least one, and most of them went to many of them!

Regardless, I'm not sure why injustice should be subject to a popular referendum. If greater than 50% of polled North Koreans praise their current political leadership, does that make their current leadership good?

Hong Kong is one of my favorite cities, and I was there in 2019 as it happens, though not for the full year. I knew people who participated in protests, and people who sniffed and said they were stupid, although when pressed, that was more out of a belief (correct, it turns out) that they would be ineffective, and were dangerous to participate in.

It is still true that something of value has been lost, while it remains a fantastic city. It is still true that multiple political prisoners are being held on charges most of us would identify as unjust.

It is also true that protesters were upset about many varied things, including the outrageous cost of housing. But it's not clear what you're trying to say with that and silly arguing about polling sources. That Jimmy Lai, Joshua Wong, and many others deserve their imprisonment? That nothing of value has been lost as China has suppressed dissent and the free press in Hong Kong?

I'm really happy that we live in a world in which people who don't know how to use the tools they have (and can't be bothered to find out) can use a coding agent to make their own tool that does what they want the way they want it to!

I'm really sad that those people keep releasing those tools to other people via cookie-cutter websites filled with misinformation instead of just enjoying them privately.

I think you're learning that Toast exists due to "skill issues."

I'm sorry Finder hurt you. Please refrain from taking it out on me.

That's progressive disclosure, and it's usually considered to be a good thing for usability.

Let's be clear: if you save something in macOS, by default you get: 1. a text field with something like "Untitled" already selected, but not the extension, so if you just start typing, the extension isn't replaced. 2. a text field for tags to make organization easier 3. a Where drop-down with recent locations, favorites, drives, etc. 4. a button with an arrow pointing down next to the "where" list. 5. depending on the app, probably a format drop-down.

Friends and family who depend on "you're good with computers, right?" people like us, they get overwhelmed by too many choices. Here, macOS gives them five, only five, and the complaint is that most people ignore four of them!

Give them the larger menu by default, and it's more options to ignore. If someone is saving every single file to their Desktop (/me shudders), then they are ignoring every bit of even a small, simple UI, so giving them a more complicated UI accomplishes nothing good.

I said they're ignoring four of the five UI elements, but now I wonder how many of the files on the desktop are named "Untitled (38).pdf" or whatever. They could be ignoring all five!

Finder is fine: not the best, not the worst.

Your list seems very preference-shaped, not pointing to actually-missing features I can see. You could make a case about discoverability, but it's hard to see how letter combinations that only work within one application and no others are easily memorized. If your habits mean you spend a lot of time working with files in a UI, then memorizing double-tap C and C+F might be easy for you. Otherwise, using the same Cmd+C and Opt+Cmd+C would have the benefit of doing what it does in nearly every other program. Or, as the other responder said, dragging auto-escapes appropriately.

I spend most of my time using a CLI, and I'm not claiming Finder is the best keyboard-driven file manager out there, but I will say that it's consistent, and works the way macOS works. Key combos work the way they work in every app, and holding down Opt updates the Edit menu to make it clear what would be different. You can do the most common things pretty easily and clearly, even if sometimes you have to use a mouse and right-click. There are also third-party apps for focused file management for pros, including some that use tuis, like ranger.

The word "most" is doing a lot here. Europe guarantees consumer rights by law, while the US relies on companies adopting the practices voluntarily. Most do, but larger companies more universally than smaller, and it's by no means universal.

It may be only a matter of time before all devs remember to append ", and make no mistakes" to the end of their LLM prompts, but I don't think we as an industry will ever reach a point at which every release of every package/library/application is scanned with the most capable model available.

I mean, we've had tooling like fuzzers available for a very long time, and most devs haven't run one against their software ever, let alone before each release.

It's the human factor I think will keep this a problem essentially-forever.

Sorry for the late response; it's been a busy week.

As a user, I don't care how complicated is the means of displaying images in a terminal session. I would only want to do so when I'm deep in a text-oriented context and there is a suddenly a need for an image. Not a chart or a graph, but an actual image. As a user, whatever contortions are necessary at that point are fine, because it's an unusual circumstance.

I hope that makes sense.

This will alarm you, then.

I set Siri to a masculine voice, because I disliked the gendered assumptions I felt with the default.

I gave my Claw Discord bot a feminine identity (Ada, with a pfp of Ada Lovelace) for the same reason. But then I set up a separate Discord bot for an LLM outside of Claw, and gave it a masculine identity so I could easily distinguish between the two mentally and expressively.

All still clankers, but "it" is too general for my dual-bot config.

Sure, of course sometimes an image conveys things better than a thousand words. But a very large percentage of what most people do with computers is primarily text, with more images in ads than useful content. By and large GUIs don't use images to convey information better, they just make text worse.

Modern terminal software supports displaying images, for what it's worth.

I use actualbudget.org to track all spending, but only update investment accounts ("off-budget" in Actual Budget terms) once a month. Completely deterministic, as all things related to numbers should be.

I have pointed my LLM at the SQLite DB and asked it to tell me what it could see from my last five years of transactions, and I was impressed with the things it picked up, and what it reminded me of, but I'm not sure I saw any value in the sense of anything I would change.

I'm going to have it review things monthly to see if that helps me, but I'm not sure it will. I'm generally already aware of how my finances are going because of my budget updates.

I kept asking this question last year, especially after that initial METR report showing people believed themselves to be faster when they were slower. Then I decided to dive in feet-first for a few weeks so that nobody could say I hadn't tried all I could.

At work, what I see happening is that tickets that would have lingered in a backlog "forever" are getting done. Ideas that would have come up in conversation but never been turned into scoped work is getting done, too. Some things are no faster at all, and some things are slower, mostly because the clankers can't be trusted and human understanding can't be sped up, or because input is needed from product team, etc. But the sorts of things that don't make it into release notes, and are never announced to customers, those are happening faster, and more of them are happening.

We review server logs, create tickets for every error message we see, and chase them down, either fixing the cause or mitigating and downgrading the error message, or however is appropriate to the issue. This was already a practice, but it used to feel like we were falling farther behind every week, as the backlog of such tickets grew longer. Most low-priority stuff, since obviously we prioritized errors based on user impact, but now remediation is so fast that we've eliminated almost the entire backlog. It's the sort of things that if we were a mobile app, would be described as "improvement and bug fixes" generically. It's a lot of quality-of-life issues for use as backend devs.

At home, I'm creating projects I don't intend for anyone outside my family to see. So far things I could theoretically have done myself, even related to things I've done myself before, but at a scale I wouldn't bother. Like a price-checker that tracks a watchlist of grocery items at nine local stores and notifies me in discord of sales on items and in categories I care about. It's a little agent posting to a discord channel that I can check before heading out for groceries.

Or several projects related to my hobbies, automating the parts I don't enjoy so much to give me more time for the parts I do. My collection of a half-dozen python scripts and three cron jobs related to those hobbies has grown to just over 20 such scripts and 14 cron jobs. Plus some that are used by an agent as part of a skill, although still scripts I can call manually, because I'll go back to cron jobs for everything if the price of tokens rises a bit more.

I was super-skeptical, and now I'm not. I think companies laying off employees are delusional or using LLMs as an excuse, but there is zero question in my mind that these things can be a huge boon to productivity for some categories of coding.

You could double or quadruple the number of pixels, and it wouldn't make any difference in how much information humans comprehend easily. You would be using more computing power and more memory to deliver the same amount of useful information less efficiently.

A "proper GUI" is rarely better than a well-designed TUI for communicating textual information, IMO. And the TUI constraints keep the failure-states for badly-designed UI tightly bound, unlike GUI constraints.

Every year or so, I look into alternative to my Paperwhite, which has been in "airplane mode" since I bought it. So far, nothing else seems to be quite up to the level of my existing device for my use case, let alone better.

It's possible I needed to log into Amazon in 2016 and 2020 when I bought my two Paperwhites, but I haven't needed to do so again since, so I'm not sure this will affect me at all. If it does, I'll have to check my notes for what was closest last year when I last checked.

Something seems wrong. A half-million tokens is almost five times larger than I allow even long-running conversations to get too. I've manually disabled the 1M context, so my limit is 200K, and I don't like it to get above 50%.

Is it... not aware of its current directory? Is its current directory not the root of your repo? Have you maybe disabled all tool use? I don't even know how I could get it to do what you're describing.

Maybe spend more time in /plan mode, so it uses tools and the Explore sub-agent to see what the current state of things is?

I appreciate these kinds of posts, and I've been semi-Degoogled for years now. What hasn't been easy for me to move away from so far is Docs/Sheets.

Apple's Numbers and Pages seem very far from ideal. Zoho didn't quite work for me. Maybe I should try Clickup.

It's easy on the creators primarily, they get to skip a lot of steps and don't have to explain away things like cell phones.

For consumers, they get quantity.

Although it shouldn't be impossible, it seems really hard to get any kind of real quality out of isekai stories. Usually the stakes are low and the setting is surprisingly similar (even though having a different setting is ostensibly the entire reason for the genre). All you end up with is variety in characters, and even then there are common tropes.

Of course! I don't disagree. And yet here we are at 69.43% desktop market share[0] for Windows for March 2023, decades after I started hearing about how this year, for sure, was going to be the Year Of Linux On The Desktop™.

Most people don't seem to be using the best possible desktop experience. Their desktop is decided by an employer, or whatever is pre-installed, or who knows what. But an ad in the start menu doesn't seem like it's likely to budge that number much.

0. https://gs.statcounter.com/os-market-share/desktop/worldwide...

Most people will sign up for a Microsoft account, or already have, and so won't even see these ads.

But let's say you use Microsoft Windows, but don't want a Microsoft account. You're going to leave Windows over this? And go where?

Wait, is 2023 the fabled Year of Linux on the Desktop, at long last?

I haven't used Windows in many years, and I'm not trying to defend Microsoft here. Still, the framing of this piece seems silly. Very rarely is any single thing reason enough for the average person to completely switch operating systems. The cost of switching is too high.

And yet Duke Nukem Forever has shipped, and Cybertruck has not. So right now, today, at this moment, DNF is not vaporware, and Cybertruck is. This is true regardless of how one feels about it.

> Regulating data privacy isn't Apple's job, if you want that fixed then you should take it up with the government or someone who can actually hold them accountable.

I live in the United States of America, where the government is bought and paid for by companies who dislike privacy for their users.

In the meantime, it may not be Apple's "job," but it's part of their value proposition, and the grumbling from software vendors indicates it's reasonably effective.

Because Facebook announced publicly that Apple's privacy policy changes would cost them $10 billion[0], putting them on the record as having strong motive to avoid the App Store.

0. https://www.cnbc.com/2022/02/02/facebook-says-apple-ios-priv...

Think about it this way: any app that refuses to go through Apple's store is telling you that you're not a customer they want. If Facebook makes that choice, it's a great time to give up Facebook!

Yes, nobody is saying that Apple Card is the only card for everyone, just that it's a great card for tap-to-pay.

4.15% is lower than 4.81%, but how did you decided that UFB defines "current market rates?"

UFB, I note, does not offer a debit card.

4.15% is not the highest yield currently available, but it would put them fourth on this list[0] of ten, making them better than average even on that rarified list. This without fees or minimums, which would put them behind only Betterment, which is not a bank, but a brokerage account.

As always with an Apple offering, there are ways that some people under some circumstances can find better terms so long as they don't care about some of the benefits Apple is offering, but that's a very long way from "below current market rates," and comes from a company a lot of people are already trusting with their funds.

It's fine if you already have an account with UFB, carry on! And next month when it's a different company leading the pack, transfer. And the month after that, while Apple is still consistently in the top five.

0. https://www.investopedia.com/best-high-yield-savings-account...

That they're blocking tracking is still privacy-focused. There are unsubstantiated claims that they exempt themselves from the same tracking, but reports of their advertising revenue doesn't contribute anything to those claims.

At Apple's scale, it's relatively easy for them to deliver $20B in ad revenue without any privacy-invading means.

People seem to have forgotten, but ads used to be based on context, so people looking at apps related to fitness might see ads related to fitness, but that wouldn't follow them around when they looked at other things. Apple still seems to be doing that; I haven't seen fitness ads on games, or game ads on fitness apps.