harrall

I don’t know about everybody else but I both sell and buy on eBay items that are more niche but not totally bespoke like on Etsy (e.g. photography equipment or specific hats).

Which means an eBay store would never make sense for my type of buyer or seller because I would just go to one of the many other existing entrenched retailers for regular things.

I know eBay pushes hard for regular retail but ultimately I see it as a marketplace for less fungible items and that’s what it excels at. Regular retail for regular items is easier because the user experience is always standard.

GameStop is trying to sit in the middle and trying to move product that is a little more fungible (previously used games and now more trading cards and collectibles) but I just don’t know how big of a market it is when you have to factor brick and mortar upkeep. I don’t want them to take down eBay in an experiment.

I actually disagree.

Someone who isn’t racist because they grew up in a progressive family just means they were lucky. They often have never been tested under pressure.

On the other hand, someone who grew up in a racist family and ends up not racist means their beliefs are battle tested. This is a real test of character — it also tells me how they process information.

What you’re describing is a third case where someone pretends to correct but has no intention to, which I do not think Bambu’s original act of opening of LAN access qualifies.

Now I think the other dimension here is that people are expecting Bambu to believe in open source. They might not actually, which is their own opinion to have, but that’s a different problem altogether. I believe in local access but not necessarily open sourcing of everything so from my PoV, Bambu’s stance is perfectly consistent.

But they complied?

Correction is much harder than starting off on the right side.

A lot of tools, community and Anthropic’s, have zero support for Bedrock.

Anthropic’s offerings for Bedrock lag behind their main platform by months, maybe up to a year or more.

New tools like LLMs probably make previously unscalable techniques scalable.

You’re not wrong but if people didn’t, all our companies would be using Oracle and Microsoft SQL Server and paying Larry Ellison instead today.

I don’t think it’s intentionality.

It’s style.

A lot of people regard technical measures as the signal of quality. The most realistic painting, the most expensive purse, the most technical flip on a skateboard, the most well drawn AI art.

It’s a cheap way to judge quality because you don’t have to understand what makes something good.

AI is really showing this divide.

I was under the impression that you delay cognitive decline (and keep brain plasticity) by learning new things.

So you need to be learning new skills, trying new sports, entering new circumstances continuously. If you’re good at something already, it’s not enough.

Employment is one of many ways of keeping things fresh because it’s easy but I see no reason why you can’t keep yourself busy too.

A transmission error does not have a strictly contained blast radius.

A bad packet could tell a flying probe to fire all thrusters on and deplete its fuel in 15 minutes.

What makes a transmission error controlled is all the protection mechanisms on top of it. An LLM cannot delete a production database unless you give it access to do it.

My hot take is that many people are naturally more comfortable with deterministic systems that have clearly analyzable outcomes. Software engineering has historically primarily been oriented around deterministic systems and it has attracted that type of thinker.

But many of us, myself included, prefer chaotic systems where you can’t fully nail down every cause and effect. The challenge of building a prediction model on top of chaos is exhilarating. I really don’t find many people like me in SWE as in, say, the graphics design department.

To me, that’s the underlying threat here — LLMs are rewriting a field that has previously self selected a certain type of person and this, quite understandably, rubs them the wrong way.

I have access to ChatCPT, Gemini, Claude and you know what I still do 99% of the time?

I open Google Search. If I want an LLM, I click “AI Mode.”

I only bother to open Gemini or ChatGPT or Claude if it’s a “big research thing” which is almost never.

An LLM is more like the unconscious part of my brain. It’s my gut. It shits out answers using an ungodly amount of parallel processing and it’s often right.

But it also hallucinates thoughts and beliefs too, and that’s where the conscious parts have to intervene.

But the conscious parts are expensive to run and I can’t multi-task that.

The conscious parts also degrade first when I don’t get enough sleep.

You can do it on Linux but you need to buy a device you attach to your monitor. I have a Spyder X Pro but there are others.

It’s like $200 and it’s not worth it unless you do color sensitive work (photo editing, printing or video editing) and you have an expensive monitor or expensive laptop with good color support. Many monitors will fail so badly the calibration won’t be able to fix it.

But if you’ve ever had a lot of trouble trying to get colors to match when printing or between devices, it could be a godsend, although it’s only one of the many reasons colors might not match.

People are freaking out about this test like it’s some judgement of their character or something. I just picked “green” or “blue” without thinking.

The biggest problem here is that people have wildly uncalibrated monitors that often have color cast tints. I color calibrate my monitors and even my factory calibrated MacBook has a slight green tint.

People should also do hue differentiation tests like this one to see if they have any color deficiency: https://www.xrite.com/hue-test

That’s way more interesting.

One of my eyes sees (very) slightly greener than the other one.

But with both eyes I got

> Your boundary is at hue 174, just like the population median. You're a true neutral.

I should test with one eye.

You can just do a mix.

Most of my devices only need 1G or even 100Mbps. No reason to switch to fiber. 1G/2.5G copper ports don’t use that much power.

For 10G+ things, it’s fiber or DAC first if possible then RJ45 if it’s the only option.

Then my backhaul between rooms is just single mode fiber, good up to 800G. Plug in a small switch at the end and you go back to RJ45 and PoE.

I only have 10G though (to transfer large files/RAWs between my computer and my storage). Something faster would be nice because NVMe SSDs can go 50G+ but that equipment is pricey and power hungry.

Google invented the AI architecture that Anthropic and OpenAI based their entire companies on? Based off years of research at Google.

Of course they should have to fight with the inventors of the technology they’re using.

There’s 100-300 earthquakes/week in California so MyShake prefers to only warn for the bigger ones.

It’s not a corner case when you have to enforce it.

Someone will end up in the middle and then you’ll be responsible for accepting or rejecting it.

The bulk is obvious but the debate isn’t for the obvious.

Well all these bugs (iTerm2’s, prompt injection, SQL injection, XSS) are one class of mistake — you sent out-of-band data in the same stream as the in-band data.

If we can get that to raise a red flag with people (and agents), people won’t be trying to put control instructions alongside user content (without considering safeguards) as much.

Google and Boston Dynamics (of Spot, Atlas fame) formed a partnership a while back and they’ve been working on building models together.

Hyundai now owns Boston Dynamics and is pushing to get the robots into their factories.

I disagree — photography has always been ripe with significant digital alteration of photos.

The main issue is that Adobe has been a long time player in the market and they have historically segmented into 4 distinct types of tools: RAW editing (Lightroom), raster editing (Photoshop), vector (Illustrator), and video editing (Premiere). Adobe still dominates in the first 3 categories.

Achieving the effects you listed would just happen in Photoshop, and Adobe never cross contaminates their product lines with the same features. You’d need to buy both Lightroom ($12/mo) and Photoshop to do what you want ($20/mo). Want vector editing? $40/mo now. Creative subscriptions are good money to them.

You’ll see other companies try to break this segmentation — for example, Affinity combined several categories of tools into one, but when they first released their suite, they actually followed Adobe’s model.

Old Fry’s had a lot of keyboards.

The Fry’s that shut down was a shell of its former shelf.

If you’re in LA, Microcenter has mechanical keyboards.

They also have a whole rotating robotic wall of 3D filament that works like a physical jukebox.

I do not think mural understanding works. It just allows you to merely accept someone, but it doesn’t mean that you want to work with them.

What America pushed after the WW2 was the “American world order” which was primarily “if we can trade, let’s forget about war and make money.” America would sit in the middle, protect shipping routes, provide a stable currency to ease trade, and encourage trade pacts.

Surprisingly, unlike beliefs, religion, language, or almost anything else, wanting to make money is… somewhat universal. It breaks down barriers. Countries wanted to work together and make money from trade. It exploited human materialistic tendencies.

But we are reaching the limits of it.

Not a US issue. A store near me has mechanical keyboards.

Definitely a some-areas-issue.

My pet opinion is that Steve Jobs was an asshole but an asshole that used his own products and used his powers of complaining to steer the whole ship to fix major "this annoys me everyday" bugs.

From my experience, "annoying but not blockers" bugs are often very neglected compared to (1) bugs that actually break things and (2) feature work. Neglecting quality of life issues leads to the "do you even use your product??" kinds of experiences.

I used to think this was true in skateboarding but eventually I found exceptions.

4-5 times / week is not a lot on its own.

You need like 20-25 hours / week. That’s how many actual hours a lot of us kids were spending, at least skateboarding.

If we take the 10,000 hour figure literally, at 20 hours/week, you get good in 9 years, which kind of fits when kids get good.

Almost zero adults I know can (or are willing to) spend 20 hours/wk on a physical hobby.

It doesn’t mean you shouldn’t learn something new, but you gotta be a little strategic.

Anduril is the biggest one. They are based in Costa Mesa, CA and are building a bigger R&D facility in "Space Beach" (Long Beach, CA) and a manufacturing facility near Columbus, Ohio.

Afterwards, you have smaller companies like Shield AI (San Diego, CA), Saronic Technologies (Austin, TX) and some other smaller ones.

Because it wasn’t planned that far. The administration probably thought it would go like Venezuela. A Middle East historian would have told you Iran has building for all our war for decades because it trusts none of its neighbors.

A second problem is that the US knew for a while that we were weak at asymmetric warfare but we didn’t fix it. There was a war game in 2002 (the Millennium Challenge, which was actually set in the Strait of Hormuz) that, though the red team did very much cheat, it did hint at a major weakness that wasn’t resolved.

There are US defense companies today that actually specialize in that but they weren’t given the same attention (but boy are they now).

This article gets ahead of itself.

The issue isn’t the splitting. There is no fiber to even split in most places. A lot of places in America had their “network” infra built 50-100 years ago on copper and no one wants to pay to basically rebuild all of it.

I happen to live in an area where there are still above ground utilities.

We got >5 gig fiber fast. We have 700Mbps 5G. I literally watched them string the fiber on the poles.

It’s still not shared, but it’s fast because it’s new. Shared would be preferred, but you need destroy + “new” first, and most people are fine with what copper gives them. Shared may even be cheaper but most people don’t think we need to rebuild anything.

Sure but I think it’s more a deeper structural problem.

When in human society have we had access to information at this rate? Yes there are bad people but there were have always and will be bad people (like people selling crock medicines 200 years ago).

We’re in this situation because it’s a brand new problem (information density) and we are still looking for a solution. I mean we now have a way to ensure medicines you buy are reasonably vetted, but it took us a while to figure that one out.

It just sucks that we have to live during the times where we haven’t figured it out, but there’s always going to some new problem grappling human society.