scott_s

Alex Ross (comic book artist who paints his pages; https://www.alexrossart.com/) also uses models for most (all?) of his work.

Johnson's whole schtick is "Don't Die", so "nation of immortals" is a logical consequence of that.

See, for example, his book he wrote: https://protocol.bryanjohnson.com/DONT-DIE-by-Zero

Or a NYT piece from earlier in the year which was heavy on that slogan: https://www.nytimes.com/2024/01/12/business/bryan-johnson-lo...

I had heat pumps installed in my house last year, to replace oil-based heat from baseboards. We have no drainage lines on the the front of the house, and only one bundle on the back. It follows the roof line, and then down a corner of the side of the house. So, yes, you will need drainage lines, but no, they shouldn't have to be all of the exterior. They should be able to be placed in an acceptable way. (Similar to, say, gutters.)

On the savings side: I think we're about even compared to oil. But I'm okay with that: we didn't have any air conditioning at all in the house, so we had no duct work. (We bought the house at the end of 2022 like this.) Since we were already doing significant work to install the duct work and get a new air conditioning system, I wanted to make them heat pumps to get off of oil. I think in the long run, we'll save money. I eventually plan on installing solar panels. And even if it's a wash in terms of money, I'm care about climate change, and I want to do what I can to electrify my house.

edit: After looking at other comments, I should clarify what we have. We have two condensers outside, and three air handlers inside. It acts like "central air", for the most part. We do have one mini-split, for the main bedroom, as we could not get ducts to it. My point here is that "heat pumps" do not necessarily mean "mini splits." My house is almost entirely centrally heated and cooled through duct work.

Outside of the northeast, most large research universities in the US are state schools.

The first three comics on this list are fantasy. At least a quarter, maybe closer to half, of the rest of the list are also fantasy.

Email the first author.

The SPEs only had individual scratchpad memory that was divorced the traditional memory hierarchy. You needed to explicitly transfer memory in and out.

If you'd like to get more attention from academics, you could try submitting a journal paper of the work. The ACM Computing Surveys journal may be a good venue: https://dl.acm.org/journal/csur

Seconded that the main reason was probably avoiding a mod. But it's not because of the compiler, but the processor. Mod and division instructions are always more expensive than additions and multiplications. If this function was on the hot path (and looking at where it's called, it's at least called once per frame), that may have mattered a lot.

Interesting - I looked into your code a bit. I found your window aggregation library [1]. You may be interested in looking into the Rust implementation of some of the research work I've been a part of [2].

In Flink, I believe the reason they need to implement their own backpressure system is that they multiplex TCP connections. That is, they have multiple logical streams flowing through a single TCP connection. If that's the case, you need to do some work to 1) detect which logical stream is the one that's blocking, and 2) don't block because other logical streams may be able to use the active TCP connection.

Thinking it through, I think what Flink's approach buys is not necessarily better performance, but better just a manageable number of connections. That is, imagine you have a process P1 with operators A, B and C. And then P2 has D, E, F. Now imagine that this is a shuffle, where A, B and C are fully connected to D, E and F. In my old system, you would have 9 TCP connections. In Flink, you will have 1.

[1] https://github.com/ArroyoSystems/arroyo/blob/master/arroyo-w... [2] https://github.com/IBM/sliding-window-aggregators/tree/maste...

I worked on a push-based streaming system which could arbitrarily distribute each operate to another compute node. Backpressure occurred naturally: if the thing you’re pushing into is busy (internal queue is full; TCP socket is busy) you wait. We had no explicit backpressure mechanism, and yet we had it in our system.

From the submitted article:

> Proteins, sauces, veggies, the dressings that go in their gigantic salads, the chicken marsala and mushrooms — it’s cooked on the spot. If there’s any part of the Cheesecake Factory that resembles an industrial machine, it’s the multiple stations and line cooks needed to create handmade food for every meal. ...

> “Their quality, their execution, and consistency across the country — it is always the same. And that’s a compliment! That’s impeccable,” pastry chef and Food Network star Young says, “The sauces, the dressings, everything is made in-house. That level of consistency doesn’t happen anywhere else.”

This is in the last section, titled "The Cheesecake Factory is a marvel." That section is the most interesting part, because it talks about the execution required to make a large amount of a wide variety of food from scratch, at scale, consistently.

Strongly disagreed. Unless you regularly buy and sell houses, when it comes to buying a house, you are out of your depth. You need the help of a professional, and you want one who is looking out for your interests.

I admit finding a real estate agent who you can trust is not easy. I don't know how to do it, other than from who you know. I was lucky, and ours was a friend of my wife's family.

I considered the emergency vehicles to be violations of the rule, but that they were defensible exceptions. To borrow a legal concept, that they are emergency vehicles was an affirmative defense, meaning I had to first admit they violated the rule.

How I see it: computer science as an academic field is roughly split between theory and systems. The theory folks tend to evaluate progress through proofs. The systems folks tend to evaluate progress through experiments.

Fascinating. I obsessively listened to Metallica as a teenager, and the Justice album in particular. I know no musical theory, but I think the thing that Lars hates about that song - the tempo changes - I love.

If you're interested in a code-walkthrough of the Linux boot process in 2004: https://www.scott-a-s.com/the-linux-boot-process-of-2004/

A fair implication, but they mean "Python-based" in that the language the compiler implements is based on Python.

Not any of the people who developed it. But some of the lay-public is under this impression, so it’s an important clarification to make.

I worked for IBM Research for a decade, from 2011 to 2021. Of that, 8.5 years were in the Yorktown Heights building mentioned in this article. The building is still in good condition, and it really is a great physical location for research.

The building is a long semi-circle, and only three stories. The exterior windows - front and back - are shared hallways. Offices and meeting rooms are in short hallways that connect the front to back. This means that offices and meeting rooms don't have windows, but I never minded that. I got plenty of natural light from the short hallway, and I loved being able to look outside when I walked around the building. And because there's effectively two main hallways per floor, you run into colleagues all the time. The private offices were small but cozy. Three people could comfortably have a conversation in someone's office; once it grew to four, you generally found a meeting room.

I used to do all of my focus work in my cozy private office. Now, I tend to do my focus work at home, and going to the office is for talking to people.

As someone who grew up in the suburbs, I liked that it was in the suburbs. But I think that's a hard sell for a lot of younger folks now. I knew people who lived in NYC and would commute up to the suburbs, but that was usually an arrangement that would last only a year or two. They would either leave the company, or stay and move up the suburbs.

It was used in the excellent show Severance: https://www.curbed.com/2022/05/apple-tv-severance-bell-labs-...

edit: I assumed they were asking about Bell Labs center in Holmedl, NJ.

I have also recently discovered his channel. It is a joy to watch him work.

It’s not cheap, but I’m happy with the OWC Thunderbolt Hub: https://eshop.macsales.com/shop/owc-thunderbolt-hub

It has its own power supply. I have a USBC 14 inch monitor and a webcam plugged into the hub, and the hub plugs into my MacBook. The hub can power my external monitor, webcam and the MacBook.

Agreed. However: powerlifting style training has more risk of injury. Pushing the limits of what your body can handle in the 1—6 rep range is riskier than pushing the limits of what your body can handle in the 6—10 rep range. With that said, I much prefer powerlifting style training.

Strange, they may have changed their text since you read it, because this is what I see:

> We read every resume. This means that a lot of tricks you’ve read about how to beat automated screening such as include certain phrases of job descriptions in your resume, repeat “hot” keywords, fill your resumes up with random metrics, etc. doesn’t work on us. They can even hurt when you apply to companies like us, because the resume space used for these tricks is the space you aren’t using for things that are relevant to us.

Emphasis mine. They're not penalizing folks for the alphabet soup, but pointing out it's a lost opportunity when you have finite space.

There are. We just have a high bar. I wrote this, but others have also pointed to it: https://news.ycombinator.com/item?id=7609289

Without a join, you are sending the entirety of Table B over the wire and processing it in Python. Then you send that back to the database, inline in a query, to do an ad-hoc join on Table A. The results are then sent back over the wire to the Python side. Notably, the result set may be small, and it may always be small. Table B may grow very large, and all of it will always be sent over the wire and processed in Python.

With a join, the database is able to do the join on Table A and Table B in place, using whatever indexes it already has built up. The only thing sent over and processed by the Python is the result set. Even if Table B becomes very large, only the result set is sent over the wire and processed in Python.

Without a join in the query, you're essentially having to replicate the kind of logic that already exists in the database engine, in Python. That is, the database engine is already doing chunking and parallelization for you.

Yes. You also have to be good enough to be able to guess what is most likely to best the best engine move.

Players only reference the engine to resolve a claim, after the move has been played.

I have difficulty with your premise, because I’ve read many pieces across many news organizations about how bad the fraud at FTX and Alameda was, and how much of a villain SBF is.