Peter Bloomfield

Unpopular Opinion: Experienced developers who '10x' themselves using AI are full of it. Sorry, I literally do not believe you. Of all the posts I've read of experienced SW developers who claim AI has made them '10x' more productive (and yes there are LOTs who literally say this) - not one has shown any evidence . The best I have seen is people who say "I finished this project in X days with AI and it WOULD HAVE taken me 10X days without AI'. Oh ? and how do you know that? One of the most important thing I have learned from 40+ years of SW development - 'we' suck at estimating how long things will take. Really Really Badly. Finally people have started doing scientific-ish studies to validate these claims. First result hot-off-the presses - 5+ year experienced SW developers given random real tasks performed 19% slower using AI then not. AND - most interesting, they believed they did it 20% faster. https://lnkd.in/gYSW4C8i

1,246

510 Comments

This is some of the really fun distributed stuff I've been working on at Helsing — a robust, bandwidth-efficient, JSON-like CRDTs for communication in highly jammed, degraded and dynamic front-line networks. I'm so excited that we've now open-sourced the core CRDT Rust implementation based on the DSON paper (https://lnkd.in/duR8NCaF), and even though there's a bunch more sitting around it to make it practical, the primitives are essential building blocks that others can hopefully also make use of! Huge props to Oliver Wangler (the data systems team lead) and the rest of the team for the effort and excellent work in getting the system production-ready 🎉 You can find it on crates.io here: https://lnkd.in/dqT6DtjY Blog post with much more details on the why, what, and how here: https://lnkd.in/dB4frJEi

414

7 Comments

If a change you made to your code broke your software, how soon would you know? One of my favourite demonstrations in my Code Craft training course is when I ask pairs to do mutation testing on code they've test-driven. Mutation testing systematically "mutates" copies of your code (e.g., turning a + into a -), line by line, and runs your tests to see if they "kill the mutant". If no tests fail after a mutation (the "mutant survives"), there's a good likelihood that line of code isn't really being tested (even if it is covered by tests). That gap in your tests suite is where bugs can creep in undetected, and eventually leak into a release. High test assurance - as distinct from coverage - is one of the self-fulfilling prophecies of TDD. If we only write code that's required to pass tests, and especially if we're in the habit of seeing that tests do indeed fail when the result's wrong, we end up in a situation where it's very hard to broke the code without any tests failing. Couple that with the micro-cycles of TDD and refactoring, where tests are being run many times an hour, and we don't have to wait long to find out after we've f***ed about :-) ------------------------------------------------------ <shameless-plug> If your team could use an injection of code craft, I specialise in training and mentoring for developers in the technical and team disciplines that enable rapid and sustained evolution of working software to meet rapidly-changing business needs. I've worked with over 3,000 developers, and hundreds of organisations all over the world, since founding Codemanship 16 years ago. (Link in profile, because algorithms.) </shameless-plug>

44

11 Comments

The amount of utter nonsense I see posted here which masquerades as 'advice' for young developers blows my mind. Yes linked in is the worst site on the internet which we all feel strangely compelled to exist on anyway, but even so... My advice to young developers is: a. Try to find something you genuinely enjoy b. Work on getting better at it c. Ignore linked in posts That's it.

181

8 Comments

I mentioned one of my “down the pub” software development opinions to James Lewis at EventCentric. He suggested that I mention it to Kent Beck and so, while I remember. But definitely, the tone here, is “crazy talk over a pint.” ———- XP optimises software development for folks with ADHD. The Extreme Tuesday Club is likely a hotbed of ADHD diagnoses. TDD, CI, Small Releases: we need those rewards to keep coming in regular doses. Pair Programming: “Stay on target, stay on target” Refactoring, Boy Scout Rule: “Ooh, while I am here, I should fix that as well. Now where was I? Oh the test list.” Sustainable Pace: We can only focus so long in one stretch, even with the above. Planning Game: one story at a time, don’t get distracted by other things. Business value, not “shiny thing.” Small stories because we get bored easily. Whole team working so that we can get to that shiny story faster. Osmotic Communication: Quick, let me blurt out my idea before I get distracted and forget I wonder if there is a correlation between XP teams achieving flow, and XP teams filled with neurodivergent people who have figured a coping mechanism for productivity.

312

67 Comments

A recruiter recently sent me CVs of Digital Forensic Imvestigators described as having been “made redundant.” The word struck me. It implies unnecessary, surplus, not needed, in reality this could not be further from the truth. But that’s not what’s happening. Companies are reshaping and realigning to meet new demands. The people affected still bring experience, skills and value, often becoming assets to the next organisation they join. As leaders, we should challenge language that diminishes and demoralises people. Words shape culture, and this one strips away dignity. What better term could we adopt that recognises the transition without devaluing the individual? #culture #digitalforensics #care

31

9 Comments

Text Corpus Comparison, Limiting Report Page access and so much more in Episode 33 of Into the Viya Verse 🔥 In this episode we deep dive a new feature released in proc textProfile that can help us compare corpuses of text, without the need for running embedding models. Understand how the new feature for limiting page access in a #SAS Visual Analytics report can unlock new ways of desiging reports and even how you can duplicate SAS Model Studio projects 🚀 The Special Shout Outs for their awesome articles on the SAS Communities go to Xavier BIZOUX, Mike Goddard and Sophia Rowland And now enjoy the episode

27

This week on Developer Voices we’ll be taking a look at Welsh Assembly. It’s an innovative spin on machine code that features instructions like lload, extra-llong integers, and a unique approach to register overflow that works by carrying a rare bit.

42

6 Comments

C#/.NET is still my favourite language/framework to develop with. While it can be a little challenging to find the latest cryptography languages, it will typically run on any system (with the .NET framework installed), and is often rock solid in its operation and capability. Here's some crypto with C# and .NET: https://lnkd.in/daEAknVQ

16

1 Comment

LinkedIn REALLY wants me to read all the posts about Vibe Coding. I get it. The challenge is this space really hasn't fundamentally changed that much since I started as a programmer 20 years ago by copy / pasting websites I'd download and then tinker with the code locally to understand how it worked. Back then, my main limitation was this: It wasn't that I didn't know what was wrong, it was that I didn't know the technical term to describe it. And without that correct terminology, I couldn't successfully research or learn via Google search. With LLMs, non-technical folks have a much better version of Google. When I started, if you didn't know the special technical term for the problem you wanted to fix, you were completely hosed. But the reason the non-technical folks will get stuck in the "LLM tarpit" (a la "The Briar Rabbit") Is that the LLM doesn't "understand" in the way that is needed for learning. It only does the pattern-matching part. I think I mostly did "just the pattern matching part" for quite a number of years with copy / pasting junk I didn't understand from Stack Overflow. So in that way, you could argue I was doing a much worse job than typical LLM code generation. So, if you're a non-technical vibe coder, congrats, you have a much better way to ramp up to becoming technical if you slow down and ask the LLM to explain the concepts (it has perfect recall and is good at this part) But your Vibe Coded app will never be "understood" by an LLM with the type of understanding that is required to actually make your app successful. If you don't understand what all the parts of your app are doing (and the LLM can help you learn if you ask!) then the LLM will delete essential features and functionality without you understanding. So welcome to the Brave New World where learning is easier than ever, and being lazy is also easier than ever. I talk about all of this with Murray Robinson in an upcoming podcast episode.

15

7 Comments

Following on from my previous video talking through the AI code generation headline, I wanted to share another take... it’s not just about AI writing code - it’s about giving tasks and tests in a safe environment. We’re seeing more tools that help junior devs and newcomers quickly generate, test, and validate code in sandboxes which is super powerful stuff. But remember, you can’t just trust AI-generated code to go live without checks. It depends on your risk level and use case. My advice? Learn to read the code yourself. If you don’t fully get it, use AI to help you understand and improve. How are you using AI to boost learning and development in your teams? #AI #CodeGeneration #DevTools #SoftwareDevelopment #NewIcon

59