HNHacker News
TopNewBestAskShowJobs

rented_mule

1,755 karma · joined July 4, 2021

submissionscomments
rented_mule··on Executions of the conceivably innocent are no better than human sacrifice
> If the author's description of the case is accurate, it really doesn't matter how you feel about the death penalty—you can replace "put to death" with "life in prison" and this is still an extreme miscarriage of justice

The difference is that you can partially undo a sentence of life in prison. The death penalty is permanent. In the absence of a perfect justice system (and perfection is not achievable in anything as complicated as a criminal justice system), innocent citizens will eventually be killed by the state with no chance of them eventually being set free if a miscarriage of justice comes to light.

rented_mule··on Litecli: CLI for SQLite Databases with auto-completion and syntax highlighting
Yes, uvx's behavior is a lot like pipx, but much faster to start up programs (just as uv is much faster than pip), whether cached or not.
rented_mule··on Litecli: CLI for SQLite Databases with auto-completion and syntax highlighting
If you install https://github.com/astral-sh/uv (it's really just two binaries... uv and uvx), then you can run many Python CLIs, and it will manage everything for you. This means resolving and downloading dependencies without touching your system Python. In this case, you'd type `uvx litecli`. If uvx can't find a suitable Python already installed, it will do its best to download a privately managed version. And it's very fast at doing all this.
rented_mule··on Moral implications of being moderately successful computer scientist and a woman
Even if you're open to seeing it, it often doesn't happen in a setting where you would be able to observe it.

A close friend of mine went to a top math/science/engineering college. A professor called her into his office to tell her that she was doing well above average, and that she should find a non-technical line of education and work. His rationale was that she had to be 10x as good as men to make it in a technical field. But she's not 10x as good, so give up now.

She went on to work on highly technical software on the campus of another elite technical college. While walking around that campus in a sweatshirt from her alma mater, a man stopped her and asked her where she got the sweatshirt. She said she graduated from that college. He laughed and said, "No, really, where did you get it?"

Another friend of mine was an intern at a FAANG and her manager said he'd make her a full-time offer if she slept with him. She didn't. On her first day at the company she ended up joining after college, she was asked out by a man more than twice her age. Given her internship experience and not knowing how powerful this man was at the new company, this was complicated to navigate.

It turns out that man was asking out every woman of East Asian descent that was single and that came into contact with him, many during their first interaction with him. That included teenage interns. He managed to get one of those interns to drink a bit too much and walk over to his apartment alone with him. Luckily, she figured out what was going on before he could get what he wanted, but the experience still resulted in significant trauma for her.

These kinds of things are still horrifically common. But all of these examples happened in relative privacy, so it's easy to not notice it if you're not on the receiving end of it and don't have friends that want to talk about it. Understandably, many don't want to talk about it, especially with coworkers.

rented_mule··on Why I self host my servers and what I've recently learned
I self-host a lot of things myself. There is one scary downside I've learned in a painful way.

A friend and I figured all this out together since we met in college in the 1980s. He hosted his stuff and I hosted mine. For example, starting in 1994, we had our own domain names and hosted our own email. Sometimes we used each other for backup (e.g., when we used to host our own DNS for our domains at home as well as for SMTP relays). We also hosted for family and some friends at the same time.

Four years ago he was diagnosed with cancer and a year later we lost him. It was hard enough to lose one of the closest friends I ever had. In his last weeks, he asked if I could figure out how to support his family and friends in migrating off the servers in his home rack and onto providers that made more sense for his family's level of technical understanding. This was not simple because I had moved 150 miles away, but of course I said yes.

Years later, that migration is close to complete, but it has been far more difficult than any of us imagined. Not because of anything technical, but because every step of it is a reminder of the loss of a dear friend. And that takes me out of the rational mindset I need to be in to migrate things smoothly and safely.

But, he did have me as a succession plan. With him gone, I don't have someone who thinks enough like me to be the same for my extended family. I'm used to thinking about things like succession plans at work, but it's an entirely new level to do it at home.

So, I still host a lot, but the requirements are much more thoroughly thought through. For example, we use Paperless-ngx to manage our documents. Now there's a cron job that rsync's the collection of PDFs to my wife's laptop every hour so that she will have our important papers if something happens to me.

Thinking carefully enough to come up with reliable backups like this makes things noticeably harder because not all solutions are as obvious and simple. And it's not something that ever occurred to us in our 20s and 30s, but our families were one tragedy away from not knowing how to access things that are important soon after we were gone (as soon as the server had trouble). There is more responsibility to this than we previously realized.

rented_mule··on Roblox is the biggest game in the world, but is unprofitable
Free cash flow and GAAP profit are not the same thing. Jeff Bezos explained Amazon's commitment to this idea 20 (and 27) years ago when many were saying that Amazon could never turn a profit, yet they were already generating enough cash to initiate big bets like AWS and Kindle... https://www.sec.gov/Archives/edgar/data/1018724/000119312505...
rented_mule··on Games People Play with Cash Flow (2020)
I am close friends with one of the founders. That friendship predates the company by several years. He did get a payout - I was with him when he pulled it out of his mailbox. A check for 32¢ for all the stock he had.

I don't know the details, but I believe they did it via a redemption rights clause in the investment agreement. https://venturecapitalcareers.com/blog/redemption-rights

rented_mule··on Games People Play with Cash Flow (2020)
We weren't in competition with Yahoo in the marketplace. But we were competing for the VCs' focus, and that's still competition.
rented_mule··on Games People Play with Cash Flow (2020)
They sold off the assets... the most valuable asset was our data. I don't know what they were paid for the data, but I'd guess a couple of million dollars given what I know they spent to migrate it to the buyer.
rented_mule··on Games People Play with Cash Flow (2020)
During the dot com boom / crash, I worked at a profitable, VC funded startup. I don't know the legal / financial mechanism by which they did it, but our VCs shut us down six months after the crash. They wanted to put all of their attention into the other company in the portfolio that survived the crash, Yahoo, because their revenue was already far higher than ours could ever be.

We were profitable (100s of K per year on revenue of a few million a year - small potatoes). Those numbers were growing steadily, even after the crash, but they weren't going to explode. We were given 30 minutes to collect our things and leave the building, but I suspect our founders knew a day or two earlier. They made a lunch reservation for all the employees that same day so we could say goodbye to each other, but we had to pay for it ourselves. It was a very strange experience.

rented_mule··on Interviewing the Interviewer: Questions to Uncover a Company's True Culture
> Interviews are serious business and if I'm in one I want to be hired

I absolutely get that sentiment, but I think it's far from unconditionally true for most people who aren't desperate for a job. Do you really want to be hired into a company where you'll be unhappy and likely to fail? I'd much rather wash out at the interview stage than be fired or leave in disgust two months later (the company doesn't want this either - hiring a candidate is typically not profitable until many months on the job). At its most productive, interviewing is about collaborating on the most accurate prediction of whether the candidate will be happy and productive in the situation being hired for. Many companies and candidates are bad at this, but I think it's the ideal we should be striving for on both sides.

I rephrase the above question, just as I do when I'm the interviewer. For example, "Tell me about a change that's been made with the support of senior leadership to improve engineering culture. What were some of the things people were unhappy about that led to that change? What is and isn't working about the change?" If the answers I get are too diplomatic, then my suspicion is that changes aren't being made to improve culture, and therefore it's headed in a bad direction (without active curation in some form, culture goes downhill). If they give me an example that had flawless results, I say "Give me an example that didn't work so well." The more transparency and introspection I see, the more hope I have for where things are headed. Nothing is black and white here, but questions like this can separate companies if you're fortunate enough to get more than one offer.

When asked well, questions like this suggest that the candidate thinks / cares about the health of the org and is willing to politely, but directly, ask tough questions to make a difference. That's leadership. It can directly increase the odds of a better offer. When I'm the interviewer and you're a candidate for a role in management or you're an IC with the word "leader" on your resume, it works against you if you don't ask probing questions like the above, because I'm expecting you to do so on the job.

rented_mule··on Massachusetts 'Reviewing' Kei Car Ban After Uproar over Canceled Registrations
I believe UTVs are illegal everywhere on paved roads and even on some government maintained dirt roads in California. But I live in a rural part of California where several of us drive them (or ATVs) around town. Occasionally someone will mention getting pulled over and warned (I haven't heard of a ticket). But you rarely see law enforcement driving around our town at all.

I've been in an even more remote part of California, where I was riding my UTV for a few hundred yards on a 55 MPH paved road to get back to my car/trailer after riding in the dirt all day. It was California State Route 182 in Mono County (population density: 4.2/sq mi). A local sheriff saw me and followed me to my car. I was certain I was getting a ticket. He started by asking where I'd ridden and what my favorite parts were. Based on that, he shared some other places nearby that I might like to ride and told me which paved roads I can use to get between them. As he left, he told me to have fun.

rented_mule··on Ask HN: Resources about math behind A/B testing
I've seen procedural errors over and over. Such errors often come down to our temptation to see numbers as objective truth that doesn't require deeper thought. In very narrowly defined scopes, that might be true? But in complicated matters, that rarely holds and it's up to us to keep ourselves honest.

For example, if an experiment runs for a while and there is no statistically significant difference between cohorts, what do we do? It's a tie, but so often the question gets asked "which cohort is 'directionally' better?" The idea is that we don't know how much better it is, but whichever is ahead must still be better. This reasoning doesn't work unless there is something unusual going on at 0 difference in a specific case. Many of us are not comfortable with the idea of a statistical tie (e.g., the 2000 US Presidential election - the outcome was within the margin of error for counting votes and there is no procedure for handling that, and irrationality ensued). So the cohort that's ahead must be better, even if it's not statistically significant, right? We don't know if it is or it's not, but declaring it so satisfies our need for simplicity and order.

Ties should be acknowledged, and tie breakers should be used and be something of value. Which cohort is easier to build future improvements upon? Which is easier to maintain? Easier to understand? Cheaper to operate? Things like that make good tie breakers. And it's worth a check that there wasn't a bug that made the cohorts have identical behavior.

Another example of a procedural mistake is shipping a cohort the moment a significant difference is seen. Take the case of changing a button location. It's possible that the new location is much better. But the first week of the experiment might show that the status quo is better. Why? Users had long ago memorized the old location and expect it to be there. Now they have to find the new location. So the new location might perform worse initially, but be better in the steady state. If you aren't thinking things like that through (e.g., "Does one cohort have a short term advantage for some reason? Is that advantage a good thing?") and move too quickly, you'll be led astray.

To bring this last one back to the math a little more closely. A core issue here is that we have to use imperfect proxy metrics. We want to measure which button location is "better". "Better" isn't quantifiable. It has side effects, so we try to measure those. Does the user click more quickly, or more often, or buy more stuff, or ...? That doesn't mean better, but we hope it is caused by better. But maybe only in the long run as outlined above. Many experiments in for-profit corporate settings would ideally be measured in infinite time horizon profit. But we can't measure or act on that, so we have to pick proxies, and proxies have issues that require careful thought and careful handling.

rented_mule··on Influential descriptions of depression are often misleading
What does it mean to be a "health individual"? If it means being completely invulnerable to any decline in health, then you're 100% correct. But only the immortal fit that description.

Soldiers going into battle are among the most physically healthy individuals on earth. And the military would tell us that they are in good mental health or they wouldn't give them control of weapons. Far too many come out of the experience with extreme mental health challenges, even if they remain physically healthy.

It's easy to imagine that one path to depression is a spiral where challenging circumstances lead to slightly weaker mental health, which increases vulnerability to challenging circumstances, repeat. It's especially easy for me to imagine because I ran down that spiral before, and for decades after, my depression set in. Part of the immense difficulty of recovery for me is the painful crawl back up that spiral, processing the old challenges as I go, all under the guidance of a trained professional.

rented_mule··on San Francisco to Ban Rent-Setting Software Amid Gouging Worry
Yep, and that's one of many parts of the supply issues in California. We bought our house 25 years ago. The house right next door is half the size of ours and just sold. Their property tax will be ~3x what ours is.

We might consider moving to a looser market, and therefore opening up one more house in a tight market, but we're retired and a large increase in property tax (plus potentially larger increases every year if it's outside California) is a serious disincentive for us.

Prop 13 is friction that resists the smoothing out of conditions across regions. It reduces market efficiency.

p.s. - It doesn't help the housing situation that we have access to incredible natural features all around us, but that's more about California's demand side.

rented_mule··on My Favorite Algorithm: Linear Time Median Finding (2018)
The metrics were about speed. And I was decades past my last internship at the time in question. But, as is so often the case, more than one of us may have been reinventing pretty similar wheels. :)
rented_mule··on My Favorite Algorithm: Linear Time Median Finding (2018)
You're absolutely right! Some learning that we came to later that isn't unrelated to what you're saying... don't just look at metrics (in the case I've described above, it was timings of operations in a large system), but look at histograms for them. You should be able to explain why they have the shape they do. The distributions are so often multimodal, and understanding the modes helps you understand a lot more nuance about your system.

We were an infra-focused team building up a scalable data handling / computation platform in preparation to apply ML at a non-trivial scale. When we finally hired some people with deep stats knowledge 10 or 12 years ago, there was a lot of great learning for everyone about how each of our areas of expertise helped the others. I wish we'd found the right stats people earlier to help us understand things like this more deeply, more quickly.

rented_mule··on My Favorite Algorithm: Linear Time Median Finding (2018)
Sorry, but I'm trying to keep this account relatively anonymous to sidestep some of my issues with being shy.

But, you're right, I was lucky to work on a bunch of fun problems. That period, in particular, was pretty amazing. I was part of a fun, collaborative team working on hard problems. And management showed a lot of trust in us. We came up with some very interesting solutions, some by skill and some by luck, that set the foundation for years of growth that came after that (both revenue growth and technical platform growth).

rented_mule··on My Favorite Algorithm: Linear Time Median Finding (2018)
I was using the term dictionary for illustration purposes. Remember, this was all in the context of MapReduce. Computation within MapReduce is built around grouping values by keys, which makes dictionaries a natural way to think about many MapReduce oriented algorithms, at least for me. The key/value pairs appear as streams of two-tuples, not as dictionaries or arrays.
rented_mule··on My Favorite Algorithm: Linear Time Median Finding (2018)
The kind of margin you indicate would have been plenty for our use cases. But, we were already processing all these log entries for multiple other purposes in a single pass (not one pass per thing computed). With this single pass approach, the median calculation could happen with the same single-pass parsing of the logs (they were JSON and that parsing was most of our cost), roughly for free.

Uniform sampling also wasn't obviously simple, at least to me. There were thousands of log files involved, coming from hundreds of computers. Any single log file only had timings from a single computer. What kind of bias would be introduced by different approaches to distributing those log files to a cluster for the median calculation? Once the solution outlined in the previous comment was identified, that seemed simpler that trying to understand if we were talking about 49-51% or 40-50%. And if it was too big a margin, restructuring our infra to allow different log file distribution algorithms would have been far more complicated.

rented_mule··on My Favorite Algorithm: Linear Time Median Finding (2018)
10-15 years ago, I found myself needing to regularly find the median of many billions of values, each parsed out of a multi-kilobyte log entry. MapReduce was what we were using for processing large amounts of data at the time. With MapReduce over that much data, you don't just want linear time, but ideally single pass, distributed across machines. Subsequent passes over much smaller amounts of data are fine.

It was a struggle until I figured out that knowledge of the precision and range of our data helped. These were timings, expressed in integer milliseconds. So they were non-negative, and I knew the 90th percentile was well under a second.

As the article mentions, finding a median typically involves something akin to sorting. With the above knowledge, bucket sort becomes available, with a slight tweak in my case. Even if the samples were floating point, the same approach could be used as long as an integer (or even fixed point) approximation that is very close to the true median is good enough, again assuming a known, relatively small range.

The idea is to build a dictionary where the keys are the timings in integer milliseconds and the values are a count of the keys' appearance in the data, i.e., a histogram of timings. The maximum timing isn't known, so to ensure the size of the dictionary doesn't get out of control, use the knowledge that the 90th percentile is well under a second and count everything over, say, 999ms in the 999ms bin. Then the dictionary will be limited to 2000 integers (keys in the range 0-999 and corresponding values) - this is the part that is different from an ordinary bucket sort. All of that is trivial to do in a single pass, even when distributed with MapReduce. Then it's easy to get the median from that dictionary / histogram.

rented_mule··on Ask HN: Can anyone from Crowdstrike explain the back story?
Not all perfect examples (e.g., some aren't metal, but they all have their own edge to them)... Kittie, Lamb of God, Dream Theater, Green Jelly, Heart, Ministry.
rented_mule··on Timezone-naive datetimes are one of the most dangerous objects in Python
In the case of ruff, I don't believe so. Your config needs to specify ALL or DTZ.
rented_mule··on Timezone-naive datetimes are one of the most dangerous objects in Python
There are linters that give you this statically, but you need to know to run the linter. An example: https://docs.astral.sh/ruff/rules/call-datetime-without-tzin...
rented_mule··on Show HN: You can program without loop and recursion and×÷- is all you need
Agreed.

> It's akin to compiling a simple c program with loops unrolled and fully static variables

Yes, and optimizing compilers take this to an extreme. It turns out that "computing" a factorial only requires my runtime language to have `mov` and `ret` instructions: https://godbolt.org/z/3Y6bGcsba

rented_mule··on An analysis of module names inside top PyPI packages
`pip install nodejs-bin` gets you node, including npm, in your venv along with bindings for calling it all from Python.
rented_mule··on Desalinating Water Is Becoming "Absurdly Cheap"
When I think of "directly", it's not just drinking, but all water usage under your direct control. That includes cooking, cleaning, bathing, flushing, watering your plants, etc. Basically everything that flows through your home. It would not include things like the water used to grow the food you buy, to water the grass in your local golf course, etc.

I'm on the board of a small California water company with ~750 customers. Our average household usage is ~100 gallons per day, a little more in the summer and a little less in the winter. Most households here are 1-2 people, mostly people over 60, and very few children. This is a forested area in the mountains (4,500 foot elevation) with no agriculture. I'm only aware of one lawn and no swimming pools. We get a lot or precipitation, so most plants and trees around here do not get watered. All that to say that we're likely lighter usage than a lot of areas, especially in California. So 100 gallons per person per day being used directly is probably a reasonable order of magnitude guess.

rented_mule··on Python with Braces
For me, what is visual clutter is all about what I'm used to. After a short period of adaptation, I found indentation guides disappeared when I was looking for them. You can also choose your styling. As I mentioned above, that was an extreme example. Here's a style that adds very little clutter (mine look a lot like this): https://i.sstatic.net/nxPrP.png

Also in the spirit of modern tooling, VSCode has had the ability to navigate GitHub Pull Requests, with all the normal code editing tooling, since 2018. I'd be surprised if it's alone in that ability in 2024. The same is possible with Gitlab Merge Requests using Gitlab's own extension. I can also use VSCode to remotely edit on almost any Linux/Mac/Windows machine that I can ssh into, or inside Docker containers.

rented_mule··on Python with Braces
Use what are often called "indentation guides". Here's an extreme form of them from the blockman extension for vscode: https://raw.githubusercontent.com/leodevbro/vscode-blockman/...
rented_mule··on Figma defaults to train AI models on personal data
Saying Google has the same rights (including publication, etc.) as you do to your book written in Google Docs is certainly over the line. I can't speak for the grandparent, but I have to assume they didn't have this extreme in mind when writing their comment. The far more interesting question is, "Where is the line?"

Is it okay for Google Docs to store your documents on Google's drives? If you want that product to exist, then it better be okay!

What if Google analyzes font usage in all docs? No docs out of millions have ever used SillyFont, so they decide to remove SillyFont from the default list to declutter the interface. I don't think that's over the line.

How about detecting that everyone is ignoring/undoing the spelling correction of a specific word and using that to realize they have a misspelling in their dictionary? I'd like that.

How about noticing that your documents all use British spellings rather than American spellings (e.g., "colour" rather than "color"), so they adapt spell checking and autocompletion? Is that over the line?

How about using everyone's text and location to build an ML classifier to determine that "go on holiday" is British and "go on a vacation" is American and using that to adapt autocompletion?

What if 99% of all occurrences of the phrase "It was a dark and" end with "stormy night", so they autocomplete that?

If you're writing your new book in Google Docs and I type the first five words, should it autocomplete the rest of your book for me? I think it's easy to agree this is over the line.

Where is the line? I suspect we won't all agree. Is it possible to come up with somewhat standardized concepts and language around this, so that vendors can publish data usage policies that let consumers make informed decisions?

← PreviousPage 7 of 12Next →