mrazomor

The article doesn't touch upon some major pain points. While the sea side is nice, it heavily relies on tourism and short term rentals. A long term rental is difficult to find during the summer time and it's not well regulated.

The capital (Zagreb) is overcrowded and its infrastructure is getting worse every year. I spent many years there. Now, I can't imagine living there again (judging by my occasional 1-2 day long visits of it).

And, I'm not sure how much the tech sector plays the role for the digital nomads. Some companies did grow a lot, but I don't find today's tech sector (mostly in Zagreb) much different from the one 1-2 decades ago.

You are mostly left to the local market, which has a significant outflow to the rest of the Europe for the past few decades.

From your experience, what's usually done when new projects start?

What I've seen, people focus on the business requirements and only the obvious risks. That's for the regular projects.

I have seen Munger's proposal in practice. In form of intentionally breaking production (in a controlled manner) to uncover unknown issues (similar to Netflix's Chaos Monkey), and in form of think tanks with senior engineers who take time to think about potential reliability issues.

What Munger describes is exactly the latter approach. From my experience, it's done very rarely. And never in some companies. While Munger advocates to do it for every project.

I don't see it that way.

Munger gives a tool to identify new (yet unknown) risks. Seeking what needs to be addressed so that the project doesn't fail.

The other thing is a tool to help us deal with the known risks (derisking the project, identifying the complete blockers). Checking if the project is doomed, and should not receive more investment.

I see it as two very different things from the project management perspective.

I don't interpret it the same way as you.

My understanding of Munger's intent: find the issues and risks that you are not aware of, by looking at the project from a different perspective.

What Google X (and many others) do: do the most risky parts first, so you fail fast and minimize the wasted effort (save time and money by not doing the less risky parts first, in case the project is doomed).

I lived in places where renting is unregulated/not protected, and in the places where it's well regulated.

IMO, being able to do a long term and safe rent is the biggest win for the financial planning and personal finance you can get:

- people can become independent as soon as they find their first job (living with a roommate is an option to get a bigger place -- valid even for decades if that fits ones lifestyle)

- high personal wealth liquidity (invest in financial markets, move to a cheaper place if the situation asks for it, move to a bigger place if the family grows)

- if the place that costs 2M to buy, an be rented for 3k, that's 50y of rent! (or even more, as the maintenance, tax & mortgage cost outperform the rent increase, not to mention a loss from not investing that money) and, think how the place will look after 50y...

But the renting must be very well regulated -- to make it a comfortable living. That's something worth (politically) fighting for.

(a happy renter)

I really liked https://blog.tartanllama.xyz/writing-a-linux-debugger-setup/

I learned a lot.

This is exciting!

BTW. if you want to know the applications of this work, have a look at this ACM SIGPLAN Keynote: https://youtu.be/JTU3JYp3JYc?si=jOZz611ATQar3Gec (helped me understand DNA more than all biology classes at my high school)

Because Google nor any other company is rich enough to use LLMs at that scale. Consider the quality vs. cost per query & latency trade off.

Two parts. The government & organic size issues. Both contributing xx%.

Any high churn change is difficult to get out at that scale. Now add all corp-fear because you are under lense of all governments. You can't keep the startup mentality with that kind of environment.

Google had LLMs before OpenAI. It also produced the core techniques for making it happen.

My take is that it was taken by its own weight. The government fines & incorporation that started to fear those fines slowed down the progress. How difficult is to launch a change in that environment? Same for the baseline -- people react to changes. You consider the baseline to be the Golden standard. How difficult is to launch a radical change considering that?

IMO, the only way out is to shrink the product. Split, shard, whatever. Be small enough compared to the rest so you are not the default target (Coca Cola, Procter & Gamble, etc. do it right).

There's something that the Netherlands, Belgium, Norway, Switzerland and Sweden do well when it comes to English. Almost all people I met from those countries spoke fluent English with good pronunciation.

What do you believe Dutch schools do well when it comes to learning English? How large are the classes? Are there any things outside the school that might play important role in learning English?

Could it be just the language compatibility? (not sure how Switzerland would fit)

From my experience, this makes sense (at least for the lower classes). I had English classes since my first grade. Considering the results, it was mostly waste of time. None of us had even basic proficiency even after years of attending English classes.

Language learning requires frequent classes and a decent amount of exposure (TV doesn't provide it). This is possible with older kids, not the young ones.

Probably because YouTube =! Google Search, while YouTube is still a subset of Google. So, going an extra mile for YouTube and not for others might put Google Search in anti competition issues.

Then again, I also find it absurd. YouTube is one of the most valuable parts of the Internet. And its lack of searchability is criminal. At least the YT search itself should make up for it. It's shame it doesn't.

That's something that Microsoft research wrote two decades ago. And those results were well known in the NLP community.

Example: https://www.microsoft.com/en-us/research/wp-content/uploads/... (Michele Banko published a few similar papers on that topic)

Err... as much I'd like to have a VR headset and use it, I simply can't.

I don't have time to immerse myself in the environment where I don't see the reality that surrounds me (as in, people which share my public transport, cars that stop at the traffic lights next to me, my kids running around me, people I work with, etc.). It's not that I need the reality, but I can't disconnect safely. Facebook is easy. Just reach the pocket, get the phone, a few swipes and you are using it. Be it in the office, around your kids, in the waiting room, ...

I don't see how VR can provide you a FB scale of users.

Typst still has a syntax that one needs to learn (or Google it).

Overleaf's slowness is not a LaTeX issue. I was using LaTeX IDEs with near-instant previews more than 10y ago.

Then again, why such emphasis on the preview? Both LaTeX and Typst are not WYSIWYG. The content is already in front of you. There's space for improvement, but building new systems is not the most optimal solution.

LaTeX is simple if you:

- have a good template (most of the conferences & some other parts of academia provide this),

- use a decent IDE (many LaTeX editors I tried in the past were great, now I'm fine with just vim),

- stick with the basics (text, tables, formulas, images; it's easy to Google it).

The last point is what matters the most. As soon as you want to change the margins, stylize the table in a different way, tune the text so it looks exactly as you imagine, is when things get complicated. People expect M$ Word and get frustrated with LaTeX. From what I see, Typst solves this in a different way -- it doesn't provide the ability to do that (yet, when it provides it, it will become as complicated as LaTeX).

Why even learn LaTeX or TeX? The Internet is full of recepies. That all you need. Why learning new syntax for formulas when you have a mountain of resources about LaTeX formulas to typeset anything you need.

There are a few "solved problems" in computer science, and typesetting documents is one of them. The progress is made by building TeXmaker, TeXstudio, TeX Live, MiKTeX, XeTeX, LuaTeX, etc. Not segmenting the space.

I agree with all said, beside the last two paragraphs. But I'd rather not debate the economical situation.

How do you see IT from the other perspectives? For example, from the accounting perspective: the profitability of the companies, the cost/benefit ratio, profit per employee, etc.

How the other industries are holding up? Sports? Music? Shows & Movies? Marketing? Law?

We are fortunate that we made the stack very complicated. And the stakeholders believe us that it's organic. Once the stakeholders become IT-proficient, we are doomed (fortunately, as the things stand, that's about never)

Finance: aim to be broke when you die.

It changed my view at the personal finance, and made some decisions much easier. If I die at the old age, my kids will be in their 40's-60's. My savings won't make a significant impact on their personal finance (assuming all is well; validated over the multiple inheritance splits within my family).

Which means, don't (extensively) save for the retirement, but invest in yourself and your children. Makes spending money for the classes, better schools, moving to the neighborhood with more options, etc. much easier decision to make.

He doesn't say it's going to be a punishment or a reward :). I'd bet on the latter. Lower operational cost! Increased stock value! Beefy bonus! Sundar takes all...

The Mathematical Experience -- https://www.goodreads.com/book/show/1113522.The_Mathematical...

It's touching many areas. For some it explains how they were developed or the controversy around them (e.g. the definition and use of infinity).

> They are not judged by the quality of their code, but by the results their code produces.

How much are software engineers judged by the code quality? And how much by the end results (shipped features/projects)?

There's a large gap between good code quality (i.e. keep the soft. engs. happy) and passable maintenance (i.e. features and bugfixes are not terribly late -- keep the execs/business happy).

When it comes to Fortran, I wish it long life. It found a great fit and does its job well. Numerical math has many peculiarities. I'm happy we have stable libraries for it. Long live Fortran!

The benefit of formal education (beside easier visa, easier access to the hiring pipelines and meeting new like-minded people) are the tests.

I was struck so many times thinking I understood something (either through lectures or self-learning) to be proved wrong by the exams or talking to other people who studied the same.

It's no wonder that TAOCP is mostly about exercises. But, the algorithms are easy to self-test for understanding and knowledge. The other subjects -- not so much. Formal education helps here.

What's the ChatGPT traffic volume (requests per second), and what's Assistant's? Or any of the prominent search engines? What would be cost of running ChatGPT at that traffic volume & comparable (user friendly) latency or reliability?

How old is the data that ChatGPT is serving? How costly or feasible is to update it?

In short, the ChatGPT is frozen in all directions. Those are technical limitations.

ChatGPT is an amazing tech demo. It distrupted the Big Tech. That's great. But, we'll need to wait for the new generation. Not ChatGPT iterative improvement.

My impression is that the past predictions were based on Linux desktop winning from the tech & UX side. That's unlikely to happen.

I'm betting on the Windows regression. MS is making it happen. Forced reboots, lengthy updates, forcing people to create Microsoft Account, ads and app store. It's horror. I don't understand how people allow that. In addition, MS Office and similar often change the whole UI or the formats. How the people/companies keep up with that?

MS has a limited amount of credit. It will expire eventually. If MS makes a bad move, it might kick off an avalanche. People leaving FAANG etc., might kick start new companies that help improving support, breaking the old relations, and building new ones.

And ChatGPT won't take off, IMO. It's a nice piece of tech, but lacks what's needed for productionisation. I didn't even include it in 2023 prediction, as 2023 won't even properly start before people forget it.

- Tech jobs at FAANG no longer look attractive or stable. Google hits the final nail in the coffin (not necessarily involving the layoffs)

- The age of smart assistants is over

- Microsoft will make a huge mistake involving Windows -> governments and companies start making concrete transition to MacOS and Linux desktop (start in late 2023)

- The all of above leads to a revolution of smaller companies, their boom, innovation, and culture which more resemble 90s than 2020s.

- Emergence of a new UI (non-smartphone/desktop/voice/VR)

- Countries start to move away from the fossil fuels for generating electricity. New nuclear revolution.

- New major Internet fragmentation

Well said! Thanks for calling it out.

> just because we're paid well.

People might try another perspective. Does $200k a year sound good? How about $100k a year?

How many say "yes" to the first, but "no" to the second? While still working weekends and staying late -- that's lying to yourself.

Working unpaid overtime, devaluates the employee. If it comes from more senior colleagues or managers, then it's at risk of creating an implicit culture. One not suitable for people with responsibilities outside the work place.

There's "Feedback" link for both the search results and the Knowledge Panel. I doubt that a large volume of similarly targeted feedback would go unnoticed (why would they keep this feature then?).

The ecosystem changed since then. Now updates of your binary's environment are more frequent, and often enforced. How often did you update or patch your Windows 98 or NT?

Today, it's false to assume that fire and forget releasing will work even for standalone Windows binaries.