alain94040

Actually, I believe he is probably correct in that case. He is discussing whether the mean is meaningful depending on the shape of the distribution. It's probably not fair to summarize his explanation as saying he doesn't understand mean and average.

HN's guidelines recommend to assume the best.

Protection was partially hacked with: https://pacmanattack.com

Maybe an idea: Could I use this to put a secret key in the gates, with some mechanism to prevent brute-force breaking? So my own personal Yubikey essentially.

Not sure how to protect against replay, with only 200 gates...

What can one do with 200 gates? I struggle to think of anything interesting.

Copilot doesn't change the fact that trivial questions (such as 'write a Fibonacci function') can already be googled.

What I don't know (I can't try the 'free' Copilot without providing a credit card), is how well it understands unusual constraints and separate pieces of C++ classes that make the real programs we write.

For instance, a simple test would be: "write a fibonacci function but skip the number 5". Does it do the right thing?

Another challenge would be a C++ class with several fields. That's what our real programs look like: composite data objects, mixing several different algorithms together to implement complex behavior.

Maybe my data structure has a hash table of items, as well as a direct link to the largest item. When I say: "write the function to insert a new item in the list, and remember to update the largest item if it is larger than the current one", would Copilot do the right thing? Each step is easy in itself (adding an element to a hash, comparing an item to another one).

Think of each transaction as a movement: it has a source and a destination. Therefore, you need to account for it in two places: where it came from, and where it went.

Does that make more sense?

I'm usually a big fan of stratechery, but for once I disagree with the explanation of Intel's decline.

I think it's much simpler: being number one in fab technology is mostly driven by volume: whoever has the most volume wins. During the PC era, Intel had the most volume, so they also had the best fabs. It's a winner-take-all, self-reinforcing ecosystem.

But in the last 10 years, mobile has had the highest volume. That's what allows TSMC to have the best fabs, and Intel is struggling. That's all there is to it. Nothing to do with services or software.

Your math makes no sense. The fewer the employees laid off, the more the dollar per employee will be. What does that mean? If a company lays off only one employee, then all of the CEO's salary could have gone to one person?

Hypercard was very much a presentation software. It had the best animations at the time, compared to Powerpoint. And it supported features that seem obvious once you hear about them: from any slide (card), you could click on an element of the slide, and it would take you to another slide about that element. Powerpoint is still mostly focused on linear "next slide" all the time.

In other words, with Powerpoint, you are presenting to a passive audience. With Hypercard, you were handing the control of the presentation to the attendee, and they could drive the presentation wherever they wanted.

Correct. Also known as Rent's rule. According to Wikipedia, it first was mentioned in the 1960s: https://en.wikipedia.org/wiki/Rent%27s_rule

The first issue could be fixed legally. Just like California makes non-compete in-enforceable, it could pass a law that says short-term moonlighting can’t be in employment agreements. This way, you could take a week off from your current job, and actually work for a week for an employer that you are interested in. Fully paid.

How many employees are already in the company? What stage of funding is the company at (seed funding, round A, B...)?

As a rule of thumb, there's about 15-20% of the entire company stock that is allocated to employees. If you are negotiating for employee #1-3, there's room. If you are employee #50, you are looking at a decimal point.

Always ask for the current valuation.

Would QA solve that problem? As in, "we replayed 1.4 trillion accident scenarios and our AI behaved correctly on all of them". Safety in numbers.

> he is making multiples of what I'm making

> Given that anything else would involve a massive pay cut

That's one solution right there: talk to your partner obviously, but it looks like you could quit the job that you hate, and could still maintain your lifestyle using your partner's income. Then find something to do that you actually enjoy. I hope your partner would be happier when you're happier.

randsinrepose has a great article on that concept, called the old guard. Definitely worth a read. How an engineer goes from an exciting new project to maintenance mode over time. It's the natural state of things.

https://randsinrepose.com/archives/the-old-guard/

Debunked, see study mentioned in https://www.bbc.com/news/world-europe-49746060

You seem to be very sure of your opinions about management's motivation. Are you yourself a manager or executive, or is this your opinion as a worker observing your employer's behavior?

It's called election 83(b). For it to work, the company grants you restricted stock instead of options. The restriction is that the company can buy back the stock on a vesting schedule, so it's effectively equivalent to stock-options. The difference is that you can buy all of them on day 1 at fair market value and pay no tax.

It's usually done for early employees who join before multiple rounds of funding occur.

You are missing the point, none of those chance meetings would have happened if they had to be scheduled online.

Figure out what you're good at and get a job doing that.

It doesn't sound like SWE is the right career for you. There are tons of opportunities for people who are tech-savvy but not hardcore developers. It's a lot more likely that one of those jobs will make you a lot happier and fulfilled.

Maybe you're good at communicating or have domain expertise. You can work in tech sales, product management, tech marketing, etc.

No need to change your mind, you are correct. I can think of one success story over the last many years where the initial development was outsourced: craigslist. Beyond that, for technology startups, development is obviously a core competency.

Here's an interesting article that starts to answer where the homeless budget is going: https://www.sfpublicpress.org/surge-in-s-f-homelessness-fund...

Here's a suggestion for really funding open source. Donations have produced minuscule results.

I think we agree the engineers want their employer to fund open source. It's up to us to establish a new normal: if your company relies on open source technologies, you are expected to give back 0.1% of your revenue to open source projects. That's it.

Obviously, the challenge is to go from today's situation, where 0 companies give 0.1% of their revenue to open source foundations, to 1. Maybe an open source startup will be the first. Then a second. Then maybe the Google engineers can pressure Google to join this program. And maybe, just maybe, in 10 years, it's become expected, and most tech companies do it.

(a) 0.1% is an example, chosen to be low enough that it doesn't impact the financial results of the company, but large enough that it dwarfs all current donations to open source.

(b) if you are worried about the complexity of donating to multiple open source projects, you can imagine that a company could donate its share in one chunk to some open source foundation, which then redistributes among individual open source projects.

I just read the answer as well. It's not "embarrassing", but it basically doesn't answer the question. Instead, it argues that the question is equivalent to asking what's the point of Python vs. C.

So in the end, the answer doesn't provide any specific answer regarding SystemVerilog and Chisel. All I found is one mention of negotiating parameters, which Verilog doesn't do. I would have loved to hear a lot more about examples of what Chisel makes more convenient than SystemVerilog.

How do you propose to design physical gates that would handle ternary logic, in a way that is more efficient than the current boolean implementation?

I don't think anyone can tell anymore whether an instruction set is superior just with theoretical high-level arguments. The only way the pros know for sure is to model the pipeline, run benchmarks and measure the resulting performance for each variation.

That movement started with Computer Architecture: A Quantitative Approach, which is from late 80s/early 90s?

True, and Google is slowly being disrupted by the next generation of Internet search (looks like disruption happens every 20 years or so).

The age where "search" is meant to return pages will eventually end. I know it's early, but you can think of GPT-3 as the next generation of search engines. You consume the entire "Internet knowledge", and you answer questions, by merging information from multiple sources. Not just returning one existing page. That's sort of what GPT-3 does, you just don't think of it as a search engine yet.

Google felt "magic" in 2000. A search-oriented GPT-3 will feel "magic" soon.

Another cognitive bias at play is that people are bad at statistics.

I'm a great driver. I make a mistake maybe once a year (and I apologize for it by waving my hand). Whereas every day on my commute to work, I see a bad driver doing something dangerous. Guess what, I drive by more than 365 cars on my commute, so my stats are actually no better than everyone else. But it feels like it.

Fully agreed. Debugging skills are not that hard to teach, while of course practice makes perfect.

Debugging is not limited to EE/CS courses of course. The sound on the TV doesn't work? But you still get a picture? So the HDMI cable is carrying the video correctly. Sound goes through the same cable, so it's not a cable issue. Do you hear sound if you switch to another input? Then the speaker is working fine. And so on. Divide and conquer.

Debugging (figuring out problems) relies on a handful of basic principles:

- having some understanding of how the system works

- looking for differences. A works, B doesn't. What are the differences between A and B?

- tracing issues through components (the data we care about makes it to the database, but the UI doesn't show it)

That's about it. Definitely teachable.

Re-posting a suggestion I made here a few weeks back:

Here's a radical idea (read to the end before you object). The inspiration is a blend of how open source projects recruit developers, and how the 20% Google side-projects used to work.

I work for company A, I'm starting to get slightly bored or interested in doing something else. Without quitting my job, I contribute a few days of code to company B, whose project I find interesting. As I contribute more to company B, I eventually quit company A and join company B full-time.

Benefits of this approach: no interviews. Know what you are getting into. Try different things and what works for you. How can we make this work: legal framework. California, if you are listening, pass a safe-harbor law that prevents companies from exclusive work arrangements, so I can always do a side-gig with company B even if they are competitors. Have company B compensate me, so we don't end up in a situation where company B is abusing job seekers by getting free work done. Maybe a law could say that I'm safe as long as my work for company B while employed by company A doesn't exceed 5% of my salary or so (so I worked for company B for a few days or weeks, not months). Would that work?