malexw

For the past 2 weeks or so I've been doing the A/B test, sending identical prompts to Fable 5 and Opus 5 to test their ability to produce design documents for new feature work. I've consistently found that Opus 5 produces more complete, accurate and "imaginative" designs than Fable, often finding design issues or nearby bugs that Fable 5 misses. However, that creativity means Opus seems to hallucinate more, while Fable's design is clearly based on the actual existing code. Or as Opus put it: "I hedged — [Fable] checked."

By pitting them against each other I get much better design work, and then I've been happy to hand off the design file to Opus 5 for implementation. But some of the assumptions Opus 5 makes leaves me wary of relying on it too strongly. This might be fixable by prompting it to ground its answers.

Oh gosh - now that you mention it, it was "Working Effectively with Legacy Code" that I was thinking of, not "Refactoring".

I think Martin Fowler's "Refactoring" might give a bit of insight here. One of my take-aways after reading that book is that the specific implementation of a function is not very important if you have tests. He argues that it can sometimes be easier to completely re-write a function than to take the time to understand it - as long as you can validate that your re-write performs the same way. This mindset lines up pretty closely with how I've been using LLMs.

If that's true, then I would think the emphasis in code review should be more on test quality and verifying that the spec is captured accurately, and as you suggest, the actual implementation is less important.

What does being founded in Montreal have to do with integrity?

> LA is currently the only city in North America building new subway lines.

That is demonstrably false. As I type this comment I can hear the sounds of excavators digging out a station for a new subway line in Toronto.

I've also done this and my record count was 120. 120 sponsored or suggested posts about things I don't care about in between the posts from people I'm actually interested in.

I'll echo what others have said - if social media is dead, it's because they killed it themselves.

Unfortunately the Midines is basically vaporware. I don't think he has actually built and delivered a cart in more than 15 years. He took my money for one about 18 years ago and never delivered.

They do occasionally show up on eBay and Reverb though.

Do you have examples of what context-oriented media like this would look like? After watching news about the pandemic, I found it interesting to see how short the collective public memory about the event was, and how little we remembered about the reasons for the decisions that were made early on. I wonder if a more context-oriented media might help people recall the sequence of decisions and their justifications at the time?

CBC is the public broadcaster in Canada, and Marketplace specifically has a good record of consumer-oriented investigative journalism like this. You might be familiar with another report a few years ago from them that found that Subway chicken may contain as much as 50% soy filler: https://www.cbc.ca/news/business/marketplace-chicken-fast-fo...

I'm personally more likely to trust Marketplace than not.

I have been using GAMMA, for Google, Apple, Meta, Microsoft, and Amazon.

"Gamma" as a word is sciencey, which I like, and swapping out Netflix for Microsoft seems reasonable to me.

I'll share a memory circa 2010 from my time at BlackBerry that comes to mind when people point out performance problems like this.

Around that time, whenever a new BlackBerry device was rebooted, it would take about 45 seconds for the OS to fully boot and become usable. I'm making up the actual times in this story because I can't remember the exact details now. Like in this video, the oldest devices would start up nearly instantly, but the boot sequence for modern devices was agonizing. Especially when you were an engineer trying to run a test that required rebooting the device!

One day, a pair of engineers finally became annoyed and curious enough to look into it and profile the problem. They found that a substantial portion of the boot time was spent parsing XML files that contained descriptions for device themes - that is, information about colors, icons, and text styles that users could choose from to customize their device. Something like a full 15 seconds was spent just on this one feature as part of the boot sequence!

So what happened? I suppose that no other engineers felt they had the responsibility or the time available to look into the long boot times until that point. But imagine being an engineer at a company like Microsoft, and one day saying to your manager or product manager: "Hey, can I spend a week on understanding why it takes a full second for Terminal to boot on Windows and see if we can speed that up?" How many managers are going to be enthusiastic about that when they're staring down a kanban board with enough tasks to fill 4 quarters?

The only reason these engineers at BlackBerry were able to look into the bootup times was because they didn't have to answer to anyone responsible for product. The team lead for this team at the time was "Engineer #2" at BlackBerry who had written all the apps for the first devices. His team was given permission to pursue whatever ideas or prototypes interested them, so they had the breathing room to chase leads like this that wouldn't necessarily pay off.

So part of the answer is that I don't think we as an industry are prioritizing performance. Once the performance is good enough, it's time to ship and move on to the next feature. But the other side is that we're building in more dynamic behaviour to our software. More customization, more flexibility, more behaviour that's defined at run-time. And this all probably makes software more useful and makes it easier for engineers to build more complex software. I think it's a similar kind of trade-off to choosing between python or Rust for a tool you're working on. What kind of performance do you need, and what kind of developer experience do you want?

I'm not sure if you're mis-remembering or I am, but the Bold 9000 (meteor) launched the year before the Storm. Though maybe you meant the 9700 (onyx)?

As I recall, the development cycle for Storm wasn't two years - it was less than a year. At that point it was the fastest cycle that RIM had ever had for a device, and it was done at the request of Verizon. Back then the iPhone was still an AT&T exclusive in the US market and Verizon demanded something from RIM so that they could compete with it.

I'll echo the fact that nobody I knew internally seemed interested in using Storm as a primary device. I think everyone knew it was junk. But Storm 2 went a long way towards a device that I would actually be interested in using.

Flight Chops is based in Canada, so anytime he's flying up here he'll fall under Transport Canada and Canadian Aviation Regulations (CARs). There are no legal requirements in Canada for pilots to wear parachutes.

Styrofoam actually is recyclable, though you'd have to check with your local municipality to see if it's accepted. For example, it's listed as a "blue bin" item in Toronto: https://www.toronto.ca/services-payments/recycling-organics-...

Has your health care practitioner suggested a probiotic of lactobacillus plantarum (specifically Lp299v)? This study published in February of 2021[1] suggests it can significantly reduce symptoms when taken daily over a multi-month period. Perhaps something to discuss with the doc if you haven't.

[1] https://pubmed.ncbi.nlm.nih.gov/33556972/

Incredible, thank you for sharing this. I live within walking distance of this museum and have a small collection of various early radio and television pieces, yet I had never heard of this place before.

> I remember this showing up in the SO sidebar a long time ago (what a time-suck that thing is).

I ended up installing this Chrome extension to block the "Hot Network Questions" parts of Stack Exchange because I found that I was getting distracted by them too often.

https://chrome.google.com/webstore/detail/stack-block/oaimjd...

A great feature not only for players, but for developers as well. If every change to game state goes through an event system you can reproduce any bug or crash as long as you're careful about making sure that the event stream is recorded before the process is killed.

The QNX-based OS started with BlackBerry 10 around 2013. Everything before that was a Java VM on top of a relatively small OS written by the company.

I would say that Jon Ronson's book "So You've Been Publicly Shamed" (mentioned in the article), is also excellent. I would especially recommend it for anyone building social media tech.

Got it. I have a project that I want to do based on a TI CC-1125 to connect 303 MHz devices in my home to a smart home system, but of course all the interesting parts these days come in QFNs. I was trying to decide whether I wanted to go through the trouble of reflowing it with hot air or if it was just worth the $100 for the booster pack development board. It's unfortunate that the economics of low-volume prototype manufacturing don't work out for the home designer.

Neat! Do you get paste stencils made as part of the PCBA manufacturing process? How many components a year were you hand soldering or reflowing before you decided to try and automate the process a bit more?

Thanks! I'm not sure how I overlooked the forums. It looks like there is someone there who has ported everything to Buster, so it looks like there is at least enough of a community to see some new software releases and hopefully offer some more pointers in how to contribute.

So this is a bit off-topic, but I pulled out my Novena this past weekend to work on an SDR project and after updating the installed packages I started wondering about the process of keeping these devices running up-to-date software and the community around it.

Are you able to offer any insight about how I, a developer with only a little embedded experience, could learn enough about the process to meaningfully contribute to keeping the Novena up-to-date? Is there still a community around the Novena?

Whenever I hear about this building I can't help but be reminded of Geoff Manaugh's idea for a Ghostbusters III, where the Ghostbusters discover that the NYNEX telephone system is actually... well, I'll let you read it: http://www.bldgblog.com/tag/ghostbusters-iii/

The Pixar Touch describes the creation of the company and first few films that Pixar made. I especially enjoyed it because of my interest in computer graphics, but it's probably a worthwhile read for most tech folks.

Creativity, Inc., is another book about Pixar that's been recommended to me, though it was recommended as a way to learn about what successful management at a tech company looks like. I haven't read it though, so I can't vouch for the quality.

I worked on a project a few years ago to build a system that used eye-tracking for context-sensitive voice commands. Imagine being able to say "zoom in" and have the computer zoom in at the place you're looking. Unfortunately the eye-tracking was never accurate enough for it to provide a good experience.

I grew up in the area and have known about the age of the trees since the mid-90s. For the people who go hiking here, it's already common knowledge.

Was that game Quake 3 by any chance? After reading about how Quake 3's event system worked (http://fabiensanglard.net/quake3/), I started using those techniques not only in C++ game engines, but even in Python GUI applications and I'm now experimenting with it in Javascript with Redux. I'm a huge fan of that pattern. It takes a bit of work to set up, but it's magical when it works correctly.

Huh. I really should have talked to you before I exercised my options last year. Clearly I didn't understand the tax laws as well as I thought I did.