bcyn

Agreed. Reading the rest of these comments are makes me feel crazy / like I’m missing something. It doesn’t sound like the interviewer was making the candidate divulge traumatic information - but rather assessing how they deal with adversity.

Very cool! Could you support sorting of some kind? Would love to at least view most recent first.

This is really awesome! Dream home project for me as well, but can't justify the cost of large e-ink displays so far (was shocked at the nearly ~$2k sticker price of that Boox Mira Pro!)

Which models perform anywhere close to Opus 4.5? In my experience none of the local models are even in the same ballpark.

I don't think the point about 401k stagnation is true. At most fee structures and optionality of funds change. How did that cost you 500k exactly?

Why? How do you draw the line between people who deserve to be "surveilled" (if you can even call it that in this case...) vs. people who don't?

You are entitled to your opinion of course but it just seems extremely arbitrary.

Great read, thanks! Could you dive a little deeper into example 2 & pre-registration? Conceptually I understand how the probability of false positives increases with the number of variants.

But how does a simple act such as "pre-registration" change anything? It's not as if observing another metric that already existed changes anything about what you experimented with.

There's no confusion about the type of edge. Just pointing out that if you are selling an edge rather than trading it yourself, you're either grifting or naive.

This doesn't make as much sense as you think it does. If you could predictably trade a flip from bearish to bullish (for example, of course there are other trend-based signals), you would not share that signal because others would overcrowd your trade (by buying/shorting and moving the price more quickly towards the trending direction than you).

A potential argument is that these signals are only applicable to a certain bracket of portfolio sizes (e.g. larger AUM funds would not be able to trade this strategy) -- but you are sharing this with folks presumably in your range of portfolio size.

It read to me as a (very) small discount for choosing to be billed annually vs. monthly.

You're correct about valuation, but the parent post was meant to address "how much liquid dollars should you expect to receive vs. 409a." You are likely to receive less in most cases (read: unless there are wildly successful public liquidity events) due to liquidation preferences.

Definitely agree with you that it's a problem that will be hard to generalize a solution for, and that the eventual solution is likely not embeddings (at least not alone).

Very interested to see what the next steps are to evolve the "retrieval" model - I strongly believe that this is where we'll see the next stepwise improvement in coding models.

Just thinking about how a human engineer approaches a problem. You don't just ingest entire relevant source files into your head's "context" -- well, maybe if your code is broken into very granular files, but often files contain a lot of irrelevant context.

Between architecture diagrams, class relationship diagrams, ASTs, and tracing codepaths through a codebase, there should intuitively be some model of "all relevant context needed to make a code change" - exciting that you all are searching for it.

Many SaaS products are tools. I'm sure when tractors were first invented, people felt that they didn't "control" it compared to directly holding shovels and manually doing the same work.

Not to say that LLMs are at the same reliability of tractors vs. manual labor, but just think that your classification of what's a tool vs. not isn't a fair argument.

LLMs are a codification of internet and written content

Only true for pre-trained foundational models without any domain-specific augmentations. A good AI tool in this space would be fine-tuned or have other mechanisms that overshadow the pre-training from internet content.

This reaction is primarily an emotional one. Why is a human rejecting a claim better than an AI rejecting a claim? Presumably the AI will one day -- if not today -- be more accurate in following decisioning logic than humans, who will continue to make human errors.

Great article, really hits a little too close to home. Definitely gives me the little extra boost in the form of a reminder that acting is always better than not acting.

Could you share some specific tactics/metrics you've defined to measure things like roadmap churn or unexpected feature demands?

This problem definitely resonates with me, and I have some very primitive ideas of tracking (i.e. tagging tickets that fall into these buckets), but I would love to learn from your experience if you're willing to share!

Why do you believe that? If the goal were to measure your individual programming skills in a vacuum, then your recommendation makes sense.

However, software engineering in a larger organization has a different goal -- it's team sport of building & operating software to serve the business purposes of that organization.

With this goal in mind, one's ability to build upon & maintain an inherited codebase is definitely relevant to success.

Look for a setting named "Preserve logs" to prevent the default behavior of clearing console/network log.

Cool, you dislike Facebook, but does that add value to the discussion about this project?

How so? Classes are extremely verbose IMO, which is the last thing I'd want in a white board interview.

It’s hard to judge out of context. IMO None meaning “disable limiting” should be a docstring/comment at the method level, describing what its return values mean, not inline next to this code.

I think you should be using `replace` more judiciously. IMO its primary use is redirecting from some kind of invalid state/URL, i.e. when the user does not want to ever go back.

For all valid states/URLs, users expect that the browser's back and forward button work as expected, and that they can save or send that link to others.

in a word, liability.

in a word, liability

This analogy falls apart because time spent to write != value to the reader.

An expert can take 5 minutes to write something that is far more valuable than an amateur can write in years. I'd also argue that it's much easier and common for amateurs to write books nowadays, so it's becoming less and less good of a heuristic to say "books > blogs, articles, etc"

I think you're agreeing with my point? Those areas are hubs for hiring too, for the same reason? Albeit not to the same degree as the Bay Area.

Would that image or media related incentive exist without the huge amount of tech work in the Bay Area? My point is that the pull of the Bay Area is cyclical - supply of tech workers fuels the SV image which fuels more supply of tech workers. I think it all started with tech workers though.

If Stanford or another incubator of tech innovators was in podunk MT, then we would be discussing that location instead.

Of course, it's not so cut and dry. It's sort of cyclical IMO:

- There are already a ton of existing technology companies in the Bay Area, and also a ton of employees.

- Students from schools around the country who are interested in tech might plan to move to the Bay Area due to the number of companies there.

- If your company doesn't hire in the Bay Area, it may miss out on the chance to hire the employees in the groups above.

The end result is an abnormally high supply of technology workers in the Bay Area. I'm not saying that's something intrinsic to the location, it seems to me it's more a result of recent circumstances.