sporkland

I went through college around 2000 in compsci and I remember even back then noticing that there were about 10% of us that cared. The other 90% were just in it for money/prestige. The dotbomb cleared them out, but post 2009 you could see the folks creeping back in due to Facebook and co.

We can agree scalpers are net negative.

And I like your ideas but I don't see why the venues and artists don't want to capture more of what people are willing to pay enabled by what the parent comment suggested.

I wonder if in your system it actually attracts fans or just people that have the time to wait for tickets.

Or a Ford f150: https://www.jalopnik.com/youre-not-wrong-american-trucks-and...

The article seemed full of weak rationale leading up to the conclusion which made me doubt the whole thing:

1. He kept citing things like PGP, C++ and distributed systems as things in common between Satoshi and Back, but would have described 75% of pragmatic comp sci folks at the time.

2. His end section about word correlations where he started with Back's weird isms and then started finding them in Satoshi's writings seemed like pure Texas sharp shooter fallacy. He started with broader scoring mechanisms and when those didn't work he started seeking out measures that fit his case better.

All of this based on a vibe of how the guy seemed in a Netflix documentary.

I have no idea if Satoshi is Back or not and would love to close this chapter. But this "reporter" seems to have started with a conclusion and then tried to find data that proved the conclusion.

I survived my Microsoft internship in 2001 by listening to Weezer blue, pinkerton, green album on a continuous loop.

I remember living pre-internet and post-internet, especially post-google and feeling like my own memory was being replaced with an Ethernet cable. The current AI models are definitely carving even more of my brain off, the only thing I'm unsure of is if I'm a better or worse cyborg at each stage. Like even with facts and data at my finger tips I still had to process decisions. I'm wondering what my bio brain's role will be as LLM's progress.

My deeply naive take is all of this stuff should be done similar to weight classes. I'm not sure what key measurable testosterone or other doping effects but measure that and separate athletes into classes based on that. Want to do steroids? Fine but you just get to compete against other dopers. As a low-T male I can go get my butt kicked by women at athletic events if I want.

Don't a lot of CEO's famously pay themselves $1 and make their wealth on equity appreciation / capital gains?

I assume a lot of people don't know the fundamental odds of something happening but have a vague sense on if Mr. Market has gone bananas or not.

Can someone help me with insights about large context models? Are there relationships that pop up at the beginning and end of long context windows that don't transitively follow from intermediate points? Is there value in the training over these longer windows vs using the more basic/closer weight distributions over different sliding windows?

So the consumer ate the tariff (I saw somewhere that they just got passed on for the most part). Now the companies are just gonna get the money back and either enrich their exec staff or shareholders?

It feels like a company should have to prove they didn't pass the tariff on to consumers in order to collect this.

You're talking abstractly but the author mentions SF and NYC in the intro where cars come with deep pain points, but at least in SF there are so many bus stops it leads to trip times that are much slower than say an Uber/lyft/waymo.

While I don't disagree with your point about Epstein case being a massive cya for a ton of people in power, the fact is that if they deeply wanted to cover up something the right way to do it would to be to actually print it and scan it, this does look like someone shortcutted some broad order to print and scan all digital media.

As a regular bike commuter in cities with hit and miss roads, With big enough tires (tyres?) it's totally fine, there are some ebikes out there with massive tires. I'd be more concerned about the blind bends and presumably due to semi-rural people driving very fast. If there's a broad shoulder or you can get off the road at a moments notice then it could be okay but as with most things the cars are the biggest danger.

As a collector of off kilter bigotry, I love finding new forms of it. Gave me a smile thinking about how much you must have nursed this grudge against morning people (they are a judgey holier than thou group on average). The particular group I harbor bigotry towards is car drivers with tinted front windows, which I find to display anti social behaviors as a group.

Not dismissing your point, but Looking at the article, it looks like it's in rust unsafe code. Which seems to me to be a point that the rest of the rust code is fine but the place where they turned off the static safety the language provides they got bit.

As told in Innovator's Dilemma mainframe companies would have told the exact same story right up until the cheaper alternative met all of their core needs. But in this case I don't think you get disrupted by copycats but instead by savvy business users building their own disposable alternatives which get just enough done.

Though I think one thing that is being overlooked is that Platforms are the hidden hero under all of this. A lot of AI products are benefiting from various cloud platforms that have been created that make it easy to deploy and opperate these apps. So as long as you are providing a sufficiently general, high productivity platform that can't be emulated by one of the major vendors you'll likely become the new runtime.

Hearing about Richard Nixon at the Bohemian Grove gave me similar vibes.

Garbage reporting: 1. AWS had an outage 2. AWS has lost a lot of employees

Conclusion: The brain drain lead to the outage...

I need an LLM trained explicitly on folks confusing correlation and causation and put a big old red dot in my address bar.

I love that there's a whole section "The talent drain evidence" trying to defend their journalistic integrity, but they then go on to totally face plant.

I've been using divvy, had no idea it isn't maintained... but I'm not sure if I'd need any more features outside of what it already does. What items do you want to see them add?

I currently use divvy and effectively a desktop per app. It's not perfect but gets me most of the way there.

As someone that has influenced company politics as an independent contributor (non-manager) repeatedly, I find you have to go through a lifecycle: 1. When you first join you need to show you can actually accomplish things, likely through code contributions / launching something. 2. After you do that you start becoming part of conversations and can start influencing directly. The key here is problem identification and bringing useful independent data. Whether we like it or not managers have built in leadership authority, IC's have to establish it by above, so usually the leadership group is dominated by managers, but if you can bring useful technical data / solutions and become a "voice of the people" if you will that lets managers not have to dig in and solve every problem you will start influencing politics.

Where I see a lot of folks fall down:

1. Focus on problems that aren't attached to either important issues: tabs vs spaces isn't gonna sink the company. If you start getting triggered about a technical thing and can't explain the impact in terms of availability, cost, productivity you'll quickly get tuned out. That isn't to say the work isn't important, it's just not workable at the political level, you just gotta fix it through influence with other IC's.

2. Inability to explain how the problem is attached to important issues: Similar to above, even if it's a real problem the ability to craft a narrative so the managers and other leaders can connect it up to real value that they'll see.

3. Discomfort with taking risks: No important problem / political problem is without risks / gaps, either in terms of timeline, impact, decision paralysis (as there are truly multiple ways to skin a cat). If you need 100% certainty to kick into gear it'll be hard to influence at the top levels as it often requires taking decisions and the inherent risks involved in signing up for those outcomes.

Before that people just took flex days and it was fine. With RTO enforcement and measurement it's made it less flexible

Thought the same, but it's much less insightful than the original.

This is the correct answer. You do a pre-filter on a permissions correlated field like this and post-filter on the results for the deeper perms checks.

If you have a field like and acl_id or some other context information on the data that is linked closely to a user's files. You can pass in the user's set of those field values to the vector database to pre-filter the results and do a permissions post check with a fairly relevant set.

The vector db definitely has to do some heavy lifting intersecting the say acl_id normal index with the nearest neighbors search but they do support it.

Does anyone have data on how much better these 1M token context models produce better results than the more limited windows alongside certain RAG implementations? Or how much better in the face of RAG the 200k vs 1M token models perform on a benchmark?

The founding fathers didn't see the corrupting influence of political parties on the separation of powers. It's covered quite well in this SMBC related comic: https://www.lawsandsausagescomic.com/comic/101

While I'm sure some of these predictions will come true and we'll have to layer some additional prompting and context engineering in here...

MCP seems to be the ultimate inheritor of ReSTful architectural practices as described by Roy fielding but instead of needing a human on the other end to interact with links (hateos) you can have an LLM bridging the gap between user intent and concrete subsequent calls.

So many issues around versioning and the fragility here will likely start to disappear. Like embeddings I'd put a bet on this being one of the major unexpected architectural improvements that lets us build our scaffolding a bit higher before it collapses to use a Steve Jobs analogy, LLM's at key glue points to create looser coupling.

The problem with this line of thought is that everyone just uses it to hate on people they already hated on. You go after rich tech monopolies, progressives generally go after big business, conservatives use it to go after the biggest business of all, the government.

We all just need to focus on monopoly in all its forms, instead of letting the politicians continue to separate us using division without difference.