Dan Luu

Files, the talk

I'm reading all of the internet comments on an old blog post I wrote on why files are difficult to use correctly (https://danluu.com/file-consistency/) and AFAICT literally every single comment that says something like "you just can do X with files and it will work" is wrong. I've only read maybe 700 comments at this point and haven't gotten to all of them, so I may find a correct comment eventually, but I don't thi...

A surprising thing (to me) about bay area groupthink

One of the things that's most surprised me about moving to the bay area is how much cultural groupthink there is. I don't mean that it exists, that was obvious when I lived on the east coast. I mean that now that I've moved here, I have no idea how people get converted into one of the numerous bay area groupthink bubbles, so it's unclear why it's so common.

An example of the kind of thing I used to see when I lived on the east coast was, a programmer in my extended social circle moved w...

Baseball scouting reports

I find baseball scouting reports from before stats were "really" used in baseball fun to read. Like any major sport, productivity is much easier to quantify than in most real-world endeavors, like programming since the game is much simpler than "real" problems are. And, among major U.S. sports, baseball is the easiest sport to quantify.

Even though many millions of dollars were on the line, rather than attempt to do this, teams decided who to draft based on scouting reports containing t...

"Caisson", the company

TIL there's a company named Caisson. The above-the-fold pitch on their website reads:

The Caisson Doublecheck API (beta) instantly and securely scans any United States-issued driver’s license and returns the data in a machine-readable JSON format, and verifies that the license data matches the embedded data, and checks that the faces are the same person.

A hiring page pitch says that this is on the blockchain.

I reall...

Interviews in practice, part 2

There's this old Tweet by Zach Holman that I think about a lot that reads "Programming interviews would be wildly different if people realized that most tech jobs are pretty boring and easy".

I agree with the Tweet and was reminded about this recently when I found out an infra team that's a cousin to my team has been trying to hire a "performance engineer" for a maybe half a year or a year now. My manager was on their last hiring panel and told me that they rejected the candidate and ...

A puzzle about prestige

There’s an obvious-in-retrospect puzzle I’ve been trying to figure out for a long time: why are so many people I know in prestigious jobs unhappy? I’m not the only person who’s noticed this: it seems like every year there’s an NYT article about someone making $500k/yr or $1m/yr who’s desperately unhappy whose friends are mostly also unhappy.

For a long time, I thought this must be some kind of sampling bias. For NYT articles, this bias is obvious — they want to tell a cert...

Dustin Curtis and Donald Knuth

When I read Dustin Curtis's old post on how you should own the best wallet, the best towels, the best flatware ($200 for a four person set), etc., because even though "[t]hese might seem like stupid things to worry about, but when you have trust in everything you own, you don’t have to worry about anything. It’s liberating and an amazing feeling.", I can't help but think of Knuth's 1974 Turing award lectu...

Algorithms interviews in practice

I've spent a decent chunk of time digging around going and finding inefficiencies at the last two companies I worked.

One thing I find funny about interviews is that a lot (most?) of the inefficiencies I find when digging around are things that I think almost all of our engineers would've gotten right if they got the thing as an interview question.

If you were to look at a global profile of all of our code when I joined a certain big company, I believe the one of the hottest f...

What's so bad about JIRA?

JIRA seems like some of the most hated software used by programmers. Right now, there's an article that's #1 on HN that's an anti-JIRA rant. Maybe half-ish of the comments agree that JIRA is terrible.

I've been using JIRA recently. I don't love it, and it has some annoying UX bugs, but it seems pretty decent? If I had to rank the bug trackers I've used, I think the ranking would be:

1. Centaur homegrown bug tracker (written by one person in a weekend)

2. JIRA

3. Bugzilla...

The fallacy of corporate intent

One of the most common fallacies I see about companies is the fallacy that the company’s actions are taken for strategic reasons. People often ask my questions like, why is company X building project B, a major endeavor costing billions of dollars? It can’t be for reasons I, II, and III. Maybe they’re playing a deeper game.

Nope. That nearly-trillion-dollar company is doing that because of a series of coincidences involving who knows whom and who was in the right place at the ...

Willful ignorance in management

One thing I’m struck by is how often VPs and execs are willfully ignorant of what it’s like to work for them. Here’s a tweet where people discuss what it’s like to work for MS and Sinofsky disagrees with someone about what it was like to work at Microsoft, under Sinofsky: https://twitter.com/stevesi/status/733699065763004416

I don’t mean to pick on Sinofsky here, I’m just using this...

What's up with Google's Spanner SLA?

https://cloud.google.com/spanner/sla

It has, what looks to me, like an unusual number of outs that allow downtime to not really count as downtime that makes you eligible for a refund. I'm not really an expert on public SLAs, but I checked the SLAs for other databases that are marketed similarly e.g., Azure's Cosmos DB, and they don't have the same funny escape hatches in definitions.

Since the SLA onl...

What's going on with Qualcomm's server team?

A few years back, when IBM was already losing a lot of hardware folks to attrition ,Qualcomm gutted IBM's POWER teams in Raleigh and Austin to start their own server chip effort. Earlier this year, Qualcomm decided that this was a dead end and decided to defund the effort. They announced this to the team relatively early this year and it made the news in May when they publisized the fact that they were looking for a buyer.

Qualcomm looked like the best hope for producing an ARM server chip wit...

Why is Bridge declining?

My partner and I started playing bridge recently, and people at the local Bridge club. People often comment on how young we are. You'll find serious competitive players of all ages, but the median age of a casual player is probably in the 60s or 70s. There are a lot of discussions about why this is. The most common reasons I hear are:

1. People who are retired have more time to play games, the reason bridge looks so old is that that's who has free time.

1.b. Bridge is...

Online games after a 20 year break

I started playing modern online video games again after a 20-year or so break (I used to play an obscure video game called "subspace" in the 90s, and would sometimes drop in for the next decade, but I basically missed what happened in modern games post quake 3).

The most striking difference between then and now is how much meaner people are in random pub/"pickup" games. My recollection from 20 years ago is that, in competitive play, there were some teams that trash talked a...

On being in it for the money

One of the most common things I hear from recruiters is that the company doesn't want people who are in it for the money. But of course I'm in it for the money. The list of books I'd like to read would take more than a lifetime to finish, as would the list of personal projects I'd like to do. I don't have a lifetime's worth of blog posts, but I it would take years to finish up all the drafts I have and I have another few years worth of posts that I haven't even started, and I expect that if I wr...

We can move as fast as any startup

I've been interviewing recently, and a lot of companies tell me how much like a startup they are, how they can move as fast as any startup, etc.

Although this is a pretty small sample size, my experience has been, the more often they tell you this, the less like a startup they actually are. Actual startups never tell me this. It would be absurd for a startup, especially an early-stage startup, to tell me that it's like a startup, so that doesn't happen. People only feel the need to say this at...

Linus's rants

Linus has been venting his spleen on LKML again. The last rant I saw was about a set of patches intended to mitigate some of the damage from the recent discovered meltdown and spectre attacks. The rant is quite popular -- it’s been on every major tech news aggregator, and there are articles about it in most of the major online tech publications (zdnet, the register, the inquirer, etc.)

If you read the comments, there’s the typical argument about how it’s not necessary to be so vitriolic ...

Internet comments

I sometimes wonder what fraction of commenters on places like HN and reddit have actually read the article. Every once in a while, and article is mis-titled such that the title doesn’t reflect the content of the article at all. When that happens, most of the comments (and sometimes all of the comments) will reflect the title and not the content. There are some biases here because articles with incorrect titles probably aren’t a random sample of all articles, but this at least indicates that ...

Information asymmetry plus monopsony hiring power

There’s an unfortunate information asymmetry in hiring, where interviewees are expected to reveal everything about themselves, submit to background checks, have references, etc., and there’s no such expectation for companies. Most won’t even give you enough information to let you do a reasonable calculation of the value of stock options. This shouldn’t be surprising, since companies have monopsony power in the labor market.

Something that’s a bit strange, if not exactly unsurprising,...

Concision vs. pre-emptive nitpicking defenses

There’s a spectrum where, on one end, you have clean writing that sounds like it’s making generalizations that might not hold, and on the other end, you have writing so packed with disclaimers that every topic turns into a book. I think Paul Graham is on the clean end of the scale. My guess is that he spends roughly 20% of the space I do to express an idea of similar complexity. Eliezer Yudkowsky is at the other end of the scale. My guess is that he spends about 10x the space I do for a give...

An unintended consequence of company propoganda

I once met an employee from Blackboard. Without thinking, I stupidly blurted out “What’s it like working on software that no one likes?” Surprisingly, rather than being offended, the person was just confused.  They thought that the software was widely loved and couldn’t believe that anyone liked the software. This struck me as pretty strange since, in my social circles, Blackboard is the most disliked software in existence. There’s more disliked software, such as Epic medical reco...

What do interview questions actually measure?

The other day, I happened to be working in the back of a room where a career panel ended up being hosted. You know, one of those things where experienced industry folks gave advice to people who were just starting their careers. I mostly try to avoid these because they tend to be all anecdote and no data, but since I was already there, I stuck around.

In the first part of the career panel, someone from [redacted prestigious company] described how they got their job. They failed the first few i...

Bootcamp vs. College

Why is it expected that bootcamps cannot cover computer science rigorously? I hear this all the time and I don't really understand the logic. One argument is that they don’t have the time, but I don’t see how that makes sense. If we look at the curriculum at my alma mater (Wisconsin), which is considered a top N comp sci school for moderate values of N, they simple don’t cover all that much material.

When I attended, schools (degrees) in the college of letters & science (which includ...

Attrition

I find that most people think it’s normal for companies to have 20% attrition per year. This seems quite odd to me and I think that 20% attrition should at least be a yellow flag.

I once worked at a company where people were mostly happy, and the attrition rate was roughly 5% a year. I think about half of that was involutionary, which puts the voluntary attrition rate as roughly 2.5% a year. That’s the rate at which people left because they wanted to change industries, or because they marr...

Relative vs. absolute

Whenever I make a relative argument, e.g., "foo is larger than bar", one of the most common responses, often the most common response, is a response about the absolute value, e.g., "wrong. bar is large".

I wonder if there's some way I can phrase arguments to avoid this problem. In my most recent post, I tried to avoid this by explicitly mentioning some incorrect absolute "rebuttals" as well as stating that I was making a relative statement and not an absolute statement. This didn't work. ...

Patreon "blog"

My plan for this blog is to use it like I'd use tumblr, if I had a tumblr. I often have thoughts that are too long for twitter but too short for my blog, so why not put them here?

One thing I've been wondering about is what happened to the strategy discussion on the board game site "boardgamegeek". There was a golden age of strategy discussion from maybe 2003 to 2006, where some of the best strategy discussion you could find on Euro games was happening on bgg. My local gaming groups would have...