mudrockbestgirl

Not my experience. I think you are either just parroting what you read online, live in the countryside, or have not been to Japan for 10+ years. I live in Tokyo and none of what you say is true here. I pay all my rent online, and have for 5 years in all places I've ever lived. I never printed out concert tickets. Seals have become less and less common and are almost never necessary anymore except in very traditional B2B settings. I have one and I have never used it despite have various Japanese bank accounts. They are officially being phased out. I don't know about the ATMs because I never have to use those since 99% of restaurants accept digital payments. Yes, there are still a few that need cash, but it's really just 1%.

Japan is definitely technologically behind. I am not arguing against that. But let's not exaggerate here.

I successfully quit Twitter. I used to be on it ~2 hours a day, now I haven't opened in for almost a year. Unfortunately that habit has been replaced by others and I still have FOMO from not being on Twitter when I see colleagues sharing Twitter links.

It's incredibly hard. The only thing that seems to work for me is to create a system to convince myself that I'm not missing out. For example, set aside a fixed 30 minutes a day, or 2 hours on a weekend, to go through the top links of your favorite social media state to satisfy the craving (I use RSS for that). That way at you can at least stop wasting time becoming distracted at random times during the day. Treat it like a cheat meal.

Another thing that has helped me cut some habits are reviews. After you spend a fixed hour on Twitter, ask yourself: What did I get out of this? Was it worth it? How did I feel before and after? Write it down in your journal. If you read your old entries week after week and they're all saying "I spent an hour on Twitter but I didn't feel like I got anything out of it" it makes it easier to quit.

Still, I can't help but feel that optimized social media feeds are the worst thing that has happened to our society in the last 10 years. Our brains haven't evolved to deal with this. And I don't think people have realized just how bad it is. Sure, we see news that social media is bad every day, but I don't think we as a society have truly understood the full implications. I think it's much worse than most think it is.

I don't understand where the bank analogies always come from. Crypto is not a bank. Holding crypto is like holding cash. If you put it under your bed and it's stolen, do you go to the government and complain about reverting the transaction?

If you want a third-party (bank) to secure your cash so you can do that with crypto, but then you're also exposed to the counterparty risk that comes with it. But that has nothing to do with crypto or fiat or whatever. If you don't trust yourself with managing the keys, put it on Coinbase or Kranken, centralized banks. If you want full control, put in your safe, i.e. manage your own key.

There are also decentralized alternatives that offer something in the middle, e.g. smart contract or social recovery wallets.

That's only true in good economic times and non-war periods and places. What you're saying is true in America right now, but not in many other countries [0], and it's also not true across time. Even in America, eventually a time will come where you can't get your money out of a bank account exactly because it's just an entry in a centralized database and because it will be in the best interest of people in power to prevent you from doing so. This, and similar restrictions, have happened several times throughout history.

The point I'm trying to make is that it's not fundamentally as different from an in-game currency as you think it is. The cycles just happen much less frequently over longer periods of time (companies go bankrupt and change faster than wars happens) so you may never see it in your lifetime if you're lucky.

[0] https://www.timesofisrael.com/lebanese-cafe-owner-hailed-for...

I guess most reasonable answer is that you should include anything that would transfer a significant amount of money to the IRS. Starbucks points and FFXIV gold may be theoretically included, but the liquidity for those is so small that the IRS just doesn't care and would never enforce it. They don't want that $100. What they're trying to do here is get a cut of the billions of dollars that come from crypto.

I know that's probably obvious to you, but I don't think there's much more to it. It's just another way to make money.

I think the op may be asking the more general question of why it is defined like this. What makes a "digital representations of value that are recorded on a cryptographically secured distributed ledger" so fundamentally different from an MMO currency that it becomes taxable? The fact that it's distributed? Why should that matter? What even is the definition of distributed? And isn't an encrypted database cryptographically secured?

That argument doesn't make much sense. You don't control your bank balance, which are just IOUs, either. It's just an entry in some centralized company's database.

I use Hugo [0] and host on Netlify, but anywhere works. Couldn't be happier. I used Wordpress and Jekyll previously, but Hugo has been a good upgrade.

The problem with Wordpress is that it's difficult to export content if you ever decide to move, unless you love dealing with SQL and format conversion. It's also annoying to host since it requires a database. Static site generators like Hugo/Jekyll can be deployed anywhere for cheap without dependencies and since it's just plan text / Markdown it's easy to move between frameworks.

[0] https://gohugo.io/

> How do I get back to who I was, in terms of feeling driven to solve problems and make impact? I feel like I’ve lost who I once was and don’t know how to get it back.

This is an interesting statement. I feel the same, but I wonder how much of the optimism and motivation I had in my 20s was unfounded and naive, stemming from a lack of experience. Having seen stuff play out multiple cycles has made me more pessimistic (realistic?) This makes it harder to become motivated, at least about technology. I guess that's a good thing to some extent, but it's also sad that I can't go back to that old naive self who blissfully worked on stuff without questioning. I think you need to deal with the fact that you can't go back to that old self because you have more knowledge that inhibits the naive optimism.

Maybe try something other than technology? How about working out? Playing an instrument? Art?

Interesting, I didn't realize this is the case. Unfortunately it looks like there won't ever be MacOS support, which makes this unusable for me.

I think my point about broader applicability still stands though. Compare the guix [0] and nix [1] homepages. The former mainly sells and operating system (and somewhat mentions a package manager), while the Nix homepage sells a tool, kind of like Docker, with concrete examples: Reproducible builds, trying new tools, declarative developer environments, docker images, cloud images. For a lot of people these are immediate practical use cases, which I think is the reason behind the adoption and hype.

I don't know anything about guix, so I'm not sure what of these you can do with guix, but at least the homepage and manual don't give me the impression that these are the use cases. I don't really need an additional package manger or a new OS, but the Nix use cases speak to me.

[0] https://guix.gnu.org/

[1] https://nixos.org/

I agree with you that the Nix language is a problem, and most likely a mistake in hindsight. Even for someone with lots of functional programming experience it's ugly with bad usability. I'm not familiar with Guix, but I am with Nix and

> Guix is an advanced distribution of the GNU operating system

This sounds like it competes (?) with NixOS, but not with Nix. A reason for the popularity of Nix is that it's not only an operating system. Most Nix users probably haven't touched NixOS, but they use Nix as a package manager, build system, development environment, etc on their local machine. That target audience is a lot larger than those who want to adopt a whole new operating system.

This adoption of Nix outside of the operating system context naturally leads to more adoption of NixOS. If you already work with Nix anyway, it's not that huge of a jump towards adopting NixOS.

I generally agree with you, but it depends on how obvious the semantics of the code are.

In your example, noNameDateFiltered is most likely superfluous. If the filter function is short it's self-descriptive what it does. Adding long variables names makes the code harder to read and adds cognitive overhead. In this case, I would strongly prefer just re-using arr or a.

On the other hand, I have seen long complex function that use single-letter variable names and end up with 20 variables that mean different things. Keeping track of what's what in such a case becomes difficult, and you want something more descriptive.

Taking a step back, the goal of variable names is to make the code easily comprehensible by someone else or you at a later time. You have a few choices to do this, in order of preference:

- The code is self-explanatory, like a simpler filter function - use a short simple variable names. This is always the preferred method. Why make it harder than it needs to be? Reading over-the-top verbose code is just as bad as the opposite.

- The code expression may cause confusion - use a more descriptive variable name, but don't go overboard

- It's difficult to describe the result succinctly even with a variable name - use comments

A few things that are nice about it once you've "nixified" your project

- Locally reproducible CI builds and tests

- Sharable developer environments/shells

- Extremely fast and efficient Docker image generation

- Composability with other projects that use Nix. It's trivial to add dependencies.

- Not dealing with VMs like you do with Docker

There is a lot more if you use NixOS and make your whole operating system functional and declarative, but I think that's more of a niche.

Why is it popular now? I'm not sure. It still has a steep learning curve (and rough edges) but over the last 1-2 years it has perhaps gotten to a point where the documentation and examples are plentiful enough that more people are willing to try it out. Also, MacOS support has improved (but still kind of sucks) and the new flake system is more intuitive than the old nix files.

The problem with the Data Scientist market is that Data Scientist is a fuzzy definition. It's pretty clear what skillset makes a good software engineer, but nearly anyone can call themselves a Data Scientist, even if they just clicking around in Excel. This has resulted in these bootcamps and certifications that "make you a Data Scientist in 14 days" - now everyone with minimal qualifications can apply for Data Scientist roles. That's why the market appears so crowded, and it is. People thought it's an easy and quick way to a high paying job, resulting in a flood of low-quality applicants for these positions.

There is still a lot of room for people who have strong engineering AND data engineering/ML/math/statistics skills. But then don't call yourself Data Scientist because that puts you into the same low-barrier camp as all the others. From my own experience it's a clear resume red flag: Almost anyone that market themselves as primarily a "Data Scientist" has little technical skills.

I'm pretty shocked by this. That someone who was involved in the fraud just as much as SBF can away with this simply by cooperating is appalling. What's the point of villainizing and heroifying the most public figures? Is that justice?

Yes, just like the FTX website is on the internet. So, do you blame the HTTP protocol for every online scam?

Anyone can put scam token on Ethereum, just like anyone can publish a website. It has no relevance to Ethereum whatsoever. What the poster meant was that the FTX, the organization, was not an Ethereum protocol, it was just a centralized financial institution that published some random token.

What you want is full transparency and incentive alignment. A government, or rather the people running the government, fundamentally don't benefit from these things because they use monetary policy to benefit themselves and the elite that is currently in control. Of course they need to get re-elected, so they must make it so that the general population believes in their "goodness" - The fed is just a group a people making decisions behind close doors with who-knows-what political incentives.

So yes, in theory it could be a centralized database but I don't think there exists this mystical "global trusted entity" that has the population's best interests as their main incentive because everything within a single nation state is clouded by politics. Banks and big tech are prime examples. That's why it needs to be decentralized, decoupled from specific governments and people in power.

The problem is the nondisclosure. When you're selling something (do this program to get a body like mine) and don't disclose that you got there by taking steroids, that's a scam. The influence may still be positive, but you're fundamentally selling something that doesn't exist or yield the results you're saying it does. So you're lying.

> Now that I’ve matured a bit, I enjoy not having much of a TV habit

I don't think you can draw the conclusion that this is because of your childhood restrictions. I have not watched TV in ~10 years, and most of my friends don't have a TV, but I have never had any restrictions. I had the TV on in the background for 8+ hours a day when I was a kid. I don't think there is any correlation at all.

> Traditional banks, for the most part, have DNA built around protecting customer interests and customer money.

I don't think we're living in the same world :) Traditional financial institutions are just as bad, if not worse, than most crypto companies. You just have look at all the financial system crashes and exchange frauds. A key difference is that the government is there to bail them out because they are tightly linked, and that traditional institutions know they'll be punished because of strict regulations. It's not about the actors, or some kind of fuzzy DNA/culture, but about government and regulation.

I think it's also important to note the discrepancy in transparency. Yes, FTX was just a centralized exchange that had little to do with the blockchain, but you were still be able to see some of FTXs balance movements on-chain, simply because they are forced to use ETH/BTC/FTT/etc. If we hadn't, FTX may have gotten away with what they're doing a lot longer. Nobody may have found out.

With traditional financial institutions you have almost no transparency. You have absolutely no idea what they're doing behind your back. All you can do is trust the government to eventually bail them out if they mess up. Or trust that they're scared enough of going to prison that they don't try shady things.

That's a great summary, but it's important to understand that much more goes into training these models. The architecture is not any kind of secret sauce, or special in any way. It's just a typical Transformer. I call this "architecture porn" - people love looking at neural net architectures and think that's the key to success. If only you know the algorithm! It's so simple!

But reality is usually much messier. The real training code will be littered with hundreds of ugly little tricks to make it work. A large part of it will be input preprocessing and data engineering, tricks to deal with exploding/vanishing gradients, monitoring, learning rate schedules and optimizer cycling, complexity for distributed training, regularization tricks, changing parts of the architecture for performance reasons (like attention), and so on.

I have no doubt that exercise is good for you, but I'm extremely skeptical about these kind of books. I haven't read this one specifically but from your description it fits this archetype of pop-science books that cite studies as proof. But studies, especially these kinds of studies, are no proof. It sounds like a huge collection of selection-bias based experiments, which already suffer from selection bias due to the academic system, spun into a narrative to sell a book to the general clueless population.

Again, I'm sure exercise is great for you (how could it be bad?), but given how difficult it is to get replication in these types of studies I wouldn't trust any of the specific results... I'd bet lots of money that, for example, "People learn vocabulary words 20 percent faster" is complete BS and not replicable.

Amazing! Added to my 1000-item list with the other "must-build habits for life success according to influencer BS clickbait posts"

Don't forget to meditate, journal, go on silent retreats, take cold showers, eat a kiwi a day, and use Rust for maximum life success!

Ah, and don't forget to breathe. I read it increases longevity.

But these are "old" models from 5+ years ago. Implementing them is not going to help you get up to speed with more recent AI research. From the OPs post, it seems like he already knows these basics.

I agree. What I meant to say is not that there won't be any use cases, but rather that they won't be as obviously transformative as people expect them to be. We won't have new kinds of products that are uniquely enabled by ChatGPT. Instead, we will get incremental improvements in spell checking, rewording text, code search, etc. But just like Gmail's autocomplete, better language translation, or better voice recognition, these will be largely behind the scenes and won't lead to big societal changes. Many of these improvements will be existing companies integrating better models, not new startups that become possible now because of ChatGPT.

I'll make an anti-prediction instead. People will fail to find decent use cases for ChatGPT/etc. We'll get used to the fact that AI models can generate language that looks indistinguishable from humans on the surface, but there won't be profound impact. GPT-3 was released in 2020, and generative image models have been decent since ~2017. In the short term there hasn't been as huge of an impact as many have predicted. It'll probably happen very slowly over the next 5-10 years. People overestimate the short term.

Twitter is going to be interesting to watch. I think it'll either crash and burn, or become huge doing something very different from what it is now.

EDIT: I'll add one more. Meta will keep declining and fails to deliver anything meaningful wrt. Metaverse.

I agree. It will finally go to zero in 2023.

It's difficult because there are hundreds of really good courses and I don't keep track. It's almost a full time job to just keep track of this stuff given how much there is. Search for one of these "awesome Deep Learning lectures" Github repos and you will find them.

The Stanford courses (224, 231, and the Transformer one) are a safe bet. In general, I would recommend university courses where the lectures are online. There are a lot of them. Fast.ai is supposed to be good as well from what I've heard, but I don't have experience with that myself.

Why are these low quality resources being upvoted? Not only is there basically no content here, but some of it requires signup. You get more than this from reading a single page of PyTorch documentation.

The web is flooded with Deep Learning tutorials that are both free and much higher quality than this. I can't tell if this is vote manipulation or if HN just upvotes anything with Deep Learning in the title without ever checking the link.

Personally I flagged this because I don't think low-quality content marketing stuff like this belongs on the front page.

You're right, I'm assuming that the interview should resemble the knowledge needed in the actual job as much as possible. Maybe there are exceptions but I think that's a pretty fair assumption. I think that's the reason why so many people hate whiteboarding interviews. They are not close to the job.