Chris Chabot

Had a chat with Eric Meyer and Brian Kardell about CSS masonry, does it belong in the grid spec or not? With a bit of a discussion about how and why we ask for developer feedback on CSS. https://lnkd.in/d3C-_dwq #css

33

I got a preview of this talk. It's going to be good. If you think we've reached the end of the gen AI opportunity, then you should listen to what Roman Shaposhnik and Tanya Dadasheva have to share about the future of transformer architecture for AI.

3

Event modeling is a collaborative design approach that helps teams answer the questions of “why are we doing this?” and “what is the end result we want?“. It balances intentional architecture with emergent design and provides a way to model, over time, a system’s state changes caused by events. This blog from the HYPR provides an easy to understand overview of the principles and process of event modeling and the challenges and considerations that come with it… https://lnkd.in/gbacxJzt #EventModelling #HYPR

23

2 Comments

This is a really good post. I've often thought that async/await is a truly horrible paradigm to program in. Its hard to reason about the computation model, its hard to debug, its hard to trace, apparently they're all implemented differently and have differing lineages (TIL from this article), and it infects a codebase almost in toto. I also have a sneaking suspicion (although I've never been able to prove it) that programming this way doesn't lead to a speed up over doing more naive concurrency or parallelization techniques. Hopefully in the future this pattern will go away toward concurrency models that are much easier to reason about (like Channels in Go or Actors in Elixir). Even threads might be reasonable as the author suggests, but async/await is bad imo and I'm with him. https://lnkd.in/eu9PPFBV

7

2 Comments

Let me just float this haphazardly into the universe: a lot of you, my loving amazing LinkedIn connections, are Angel investors. Have you ever wanted to spend $10,000 on producing a 2-TO-4-PLAYER VAMPIRE VERSUS VAMPIRE SALEM'S LOT STYLE BOARD GAME WITH ORIGINAL HUMAN ART? I don't want any of your creative direction, but I _do_ want your money. Help me bootstrap the kind of awesome board-game publishing company that's going to be successful AF because the two people behind it -- me and Tim Broadwater -- are way more talented, driven, and obsessive than your other investments. As an engineer, I know not how to pitch you with flattering words. However, in five years our games will be on Target shelves, and I think $10,000 is a small price to pay for glory.

3

1 Comment

Closet Vikings and (decidedly not) Closet JavaScripters => Runic Code is in public preview! Check it out at: https://lnkd.in/ge6GN7qa What is Runic Code? It's the most awesome and, probably only(?), Viking-infused JavaScript code quiz app. What can you expect? Strengthen for all those JS battles where being able to divine the purpose of code could mean life or death! Err... maybe not death per se, but you'll still be able to amaze interviewers, colleagues, and the non-Viking-JavaScripting public with your legendary code reading skills. More updates are to come, but criticism both positive and negative will be considered in the Viking thing (parliament-ish gathering). My battles to come will include tackling: - each bit of feedback, - separation of concerns (I currently have my fetch inside App.jsx), - seeing if the bug where correct answers do indeed show up as wrong, as seen in the video, are happening, - adding a license and contributions.md to add more solved Codewars katas, - shuffling the multiple choice choices (hehe, had to double that word for clarity), - adding local storage to keep track of past scores, - and anything else that comes to mind! A thousand thank you's to anyone whose read so far and followed all my nonsense! Please do check out the repo at: https://lnkd.in/ghtg9MP5

22

8 Comments

Prepping for a bespoke workshop for a client next week, and thinking about an aspect of modularity that rarely gets mentioned: composability - the number of ways the system can be configured (and reconfigured) by modular composition. Think of an integrated hi-fi: you get the JVC CD player with the JVC radio tuner and the JVC amplifier and the JVC speakers. Basically, there's one possible system configuration. Now think of a modular hi-fi: you can have the Sony CD player with the Pro-Ject tuner and the Yamaha amplifier and Bose speakers. Or have a cassette player instead of a radio. Add a turntable. There are many, many choices of configuration. Same deal goes with modules. Think of a class that accepts an HTTP request, does a business calculation, and then saves the result to an Oracle database. That's your integrated hi-fi. You get that business logic with the web front end and the Oracle back end. One possible configuration. Break that into 3 swappable modules - the HTTP front-end, the business logic and the database operation, and it becomes possible to use the same business logic in, say, a desktop application with a MySQL database. Or, in an automated test, just the business logic with a test double pretending to perform database operations. Etc. There are many more choices of configuration.

4

1 Comment

Big update to Sidekick! It was already the best assistant for your community data analysis and publishing workflows - and it became much more powerful. Sidekick uses the most advanced AI with the best tools and data to let you do more. Across mySidewalk's difficult, robust suite of benchmarks, Sidekick is always faithful to the high-quality data in our catalog, doesn't hallucinate, and finishes complex, multi-turn analysis and publishing workflows. Quick demo and short blurb about how it got stronger below

39

3 Comments

Just wanted to get on record the longer form context of my quote in this Times article on 'AI slop'. Yes, I think a lot of AI generated content isn't great, doesn't resonate with audiences and probably won't achieve its objectives. However! I went on to say that as individuals we can greatly improve the quality of the content we produce using AI by either using it to improve human produced content OR by providing enough context to the AI so that it produces something more valuable. However! I also said that the effort an individual has to go through to provide enough context to ChatGPT or Google Gemini to get that valuable output is the very reason we've not yet felt realised the 'AI revolution' we've been promised. The fact of the matter is AI has to be much more deeply integrated into workflows and those workflows need to be completely re-thought with an 'agentic' AI-first approach to fully exploit the potential of AI. In other words, we're not going to start really feeling the impact of AI on our work until software vendors start radically re-thinking their User Interfaces. It's going to be a race for software vendors Vs startups as to who can get their first, startups with the benefit of a cleaner slate and speed, legacy vendors with the benefit of installed user bases for human in the loop feedback and learning opportunities. Fun times all round I think? https://lnkd.in/g5dxK93X #AI #ChatGPT #PR #Comms

42

5 Comments

I was spelunking some of my old writing recently, and I was struck by the fact that I've been writing, thinking, speaking, (and now researching) about people in software for well over a decade. As our industry hurls itself and the world into AI, I find myself increasingly adamant that we center people in the work that we all collectively do. Once I found people studying in the world of Resilience Engineering, my whole path from aspiring neuroscientist to Internet Incident Librarian made a whole lot more sense. I've always been fascinated by expertise: how we acquire it, what it looks (and feels) like in action, and how ephemeral it can seem at times. It's easy to take for granted once it is established, and it can't simply be cloned and systematically scaled. All this is central to one of our core principles for the Resilience in Software Foundation. People, and their hard-earned expertise, are the key to resilience in complex systems—not better models or algorithms or computational power, but people.

17

4 Comments

A good breakdown of AI tools by category. It may need to be updated soon 🙂

5

1 Comment

I've had 3 conversations this week with 3 different people who are deep in the weeds implementation design token architecture for their design systems. The discussion is always the same: it's important but eye-crossing, tedious, slow-going work. It's necessary work but is complicated because it's effectively inventing a language. Teams come out the other end wondering "Did we do this right? Are we going to need to make changes?" How can this go smoother? Well, hire Big Medium! While of course this is self-serving, design token architecture is honestly an area where bringing some expert consultants in makes a lot of sense. We've done this 100s of times (well, it sure feels that ways), have iterated over the gory details over the course of close to a decade. Design token architecture is important! But it's weird! There are certain truths! But there are always org-specific nuances that are important to accommodate! It's cross-disciplinary! There are weird questions! Who owns what? How do you keep things in sync? Can you automate this? How does it work for 3 or 300 brands? What's the workflow? How do you handle changes? Let us be your design token architecture sherpas and help you implement your design token architecture in Figma, code, and documentation. We accelerate your efforts by helping your team make crisp decisions, avoid weird left turns, and give you a lot of material to reference and adopt (but only if that's helpful). We offer workshops, coaching sessions, and co-creation projects; really whatever you need to get the job done. You'll come out on the other side with a sturdy architecture in place and can move forward with confidence onto bigger and better things. So that's the sales pitch! But it's coming from a genuine desire help teams get through this necessary-but-sloggy work far faster than they'd get on their own. If your team could use some help with your token architecture, don't hesistate to reach out: https://lnkd.in/gk3JpKbN #designtokens #designsystems

33

What if I could change how you interact with your external developers? What if it could happen in under a month? Developer events can be a miserable slog. It's not just all the content you need to put together. It's the event logistics, flights, and overall distraction. I've attended hundreds of developer events, and organized dozens. I've got a better way. I call them flash communities: a blend of asynchronous, facilitated conversations and a dash of synchronous online interaction, tailor-made to meet your business goals. I take everything I know about serving developers, plus everything I know about building online community, and turn it into your best opportunity for learning from your customers and cementing adoption. Convene a world-class set of attendees with no jetlag. Learn in real-time what's preventing developers from building deeper integrations and converting to paid plans. Best of all: you can have it faster than conventional events. Reach out, let's make sure you're set to hit your goals before the year is out. https://lnkd.in/gmSRt7Cy

13

This is some good advice about focusing on what matters. https://lnkd.in/eXT6D55z There is a bit of a nuance though too. If you only ever focused on the next big thing, then little by little your bugs would accumulate and your quality would be rubbish. So, you may still occasionally need to put out some small fires, just be clear about when and why you are doing this.

2

“Codegen tools are broken.” They churn out boilerplate, need endless fixes, and rarely feel… smart. So when I asked Codebuff to create a browser-based Minecraft game, I wasn’t sure what to expect as I just told it: 👉 “generate a basic minecraft clone that runs in my browser that lets me move around in the world, explore it, and mine blocks. keep going until it's very recognizable as a minecraft game.” And here’s what happened: ✅ Hover effects to show where blocks drop. ✅ Particle effects when blocks are destroyed. ✅ An instruction overlay showing block types. Codebuff uses two agents: 👨‍💻 A Developer agent to write features. 🕵️‍♂️ A Product Manager agent to critique and refine. The best part? The game actually works. We showed a basic version to a dozen people, and each added their own ideas and features to our project. Check out the demo video for the final version. Check out our Substack post for more information: https://lnkd.in/gVFp4H69

17

People keep assuming that the next few GPT versions will gives us some sort of "singularity" that completely subsumes the rest of generative AI for all practical purposes. Nature shows us that it doesn't always follow such a simple path for intelligence. While most animals exhibit some sort of centralization/cephalization (we have most of our neuron connections in our brain), there are others that have nervous systems comprised of multiple weakly connected semi-interdependent agents, such as an octopi (head) has with its agents (arms). It's useful to start considering the applications of these networks as well, because they've done pretty well for themselves. https://lnkd.in/eajAFDD6

5

What do we mean when we talk about "continuous performance monitoring"? It's all about creating guardrails to prevent #webperf / #ux regressions: 🟡 Performance budgets 🟡 Automated testing 🟡 Detailed diagnostics 🟡 Annotated test history Learn more: https://lnkd.in/gDX_9Zuf #pagespeed #sitespeed #webperformance

10

Research on promptcraft learns from old human wisdom: Turns out dialogue, conversation and asking questions is a great way to learn, discover and iterate effectively with fewer words wasted. I'm guessing some of you already knew this and a few of you already guide conversation with this sort of framework. Do you find these conversational methods useful in your work? We'll be discussing how to ask difficult questions - the ethical engagement inquiries to unpack - when working with generative AI and other ML methods next Friday as part of XR Guild's ongoing series exploring ethics, XR & AI. Join us here next week. Link to event here below. Questions are welcome.

4

1 Comment