Good wastebook list this episode, but first…

Even the Germans suffer from RDD

Some people actually did a study on resume-driven development. I thought it might be a joke at first, but, no: it appears to be serious. I’m working on refreshing a paper on the pitfalls of building your own application platform (“DIY platforms”),1 so the topic came up and the paper fit in nicely, see below.

Many of the pitfalls we've discussed so far touch on strategic choices and project management challenges. Resume-driven development (RDD), however, is a particularly insidious pitfall because it stems from within the organization, often a consequence of misaligned incentives.

In its innocent form, RDD occurs when your staff, fueled by genuine curiosity and a desire to stay cutting-edge, underestimate the true costs and complexities of building and maintaining a new platform using the latest technologies. They see a fascinating new tool and believe it's the perfect fit, even if it adds unnecessary burden. In its more cynical manifestation, RDD is less about genuine interest and more about self-serving ambition: builders advocating for a platform built with trendy technologies specifically to pad their resumes with highly sought-after experience.

Regardless of its flavor, RDD carries significant organizational risk. As one study on this phenomenon notes:

Extensive RDD-based technology selection may therefore lead to complex or even unmaintainable software consisting of technologies which are not suitable for the requirements, which are unfamiliar to current or future employees, or which did not deliver on their promise and were discontinued.

Part of the blame for this incentives problem can be found in the hiring process itself. Knowing that technical talent is often drawn to novel challenges, those crafting job descriptions might, perhaps unknowingly, make roles sound more cutting-edge and trendy than they truly are. This inadvertently fuels a vicious cycle: prospective employees expect to use new technologies, they actively seek roles that promise this exposure, and once hired, they then push to build with these new technologies.

It's a natural inclination, of course. When you hire people to build systems, it should come as no surprise that they will, indeed, build systems. This tendency becomes especially salient when it comes to platforms. As new building blocks and architectural patterns emerge, these talented builders will be eager to learn their intricacies and integrate them into new platforms.

However, while there are perfectly valid reasons to adopt and even build with new technologies, the danger with RDD is that fundamental architectural and strategic decisions get made based on the individual career development of your staff, not on the long-term, strategic benefit to the organization. This inherent conflict of interest can be a costly blind spot.

When contemplating building your own platform, resume-driven development is a common pitfall that demands careful scrutiny. Be sure to look beyond the allure of the new and evaluate the true benefits of each option against the comprehensive costs and risks, many of which are outlined in the preceding sections.

Here’s a related piece I wrote in August 2024 on the topic.

WhitneyGPT

On this week’s Software Defined Interviews, I interview Whitney about her new infatuation with generative AI. She’s mostly been using it for coding, but also goes over some CFP abstract uses. Listen to it! Subscribe! Enjoy!

Wastebook

• “Did I ever tell you the time I actually slipped on a banana peel? That was the moment I realized my life is a comedy, not a tragedy.” Guest DJ, The Lounge Show, May 31st, 2025.

• “Loudcasting.” Hate them so much.

• “I’m not the Russell Davies anyone wants.” The other Russell.

• “I’d be a lot more into this stuff if it wasn’t being steamrolled by a handful of tech companies, off the back of exploitation and theft.” Worth pondering for AI marketers.

• “a sudden explosion of storefronts selling every imaginable commodity, bottling water and branding bread.” Reviewing Gianny Rodari.

• ”went to columbia university film school but I only stayed there a single semester. I had some wonderful professors but I learned that I don’t like having a lot of conversations in order to work on a story.” Eric Chase Anderson.

• And: “I put a map at the start of every single chapter instead of just at the start of the book. I assumed this would blow the collective minds of the world’s eleven year olds. however, in a frustrating turn of fate the book was never marketed to the eleven year olds I’d made it for.”

• “Even Gwyneth Paltrow eats bread now.” Warren Ellis.

• “If there’s one thing people with depression are used to hearing – it’s that other people have it worse.” Internet Commentary.

• Competitive outlier.

• “Executize this” -> “write this for someone who doesn’t care about the topic but needs to deal with it for their job.”

• “The Robot Chicken sphere.” Blank Check.

• The AI Fig Leaf - If you consider great literature and art - and entertainment - AI could have created none of it, because it’s so censored and “mid.” There are no boobs and dicks, shits or fucks. Perhaps a place for humans to establish competitive advantage.

• The purpose of a to do list is what gets done. Wink, wink.

• “I don’t need this. But I’ll never throw it away. It’ll be someone else’s problem one day.” Noah.

• Theoretic parking spot in Brussels hotel garage.

• “a neurasthenic who, in the last three years of his life, locked himself up in a cork-lined apartment, ate an unbalanced diet of coffee and 1 - 2 croissants a day, and became largely nocturnal.” Proust.

• And: “One easy way to distinguish true art from mere entertainment: How many years did the creator spend languishing in bed to produce it?”

• “‘I’m not capable of the work I want to do. How can I make better work?’ 'Take something you like,” she said, ‘and try to copy it exactly. Copying teaches you a lot about technique.’" Ibid.

• “Not my fondest memory.”

• ”on Twitter or Bluesky, there are five major varieties of short-form social-media post: Here is some information; Look at how funny I am; Look at how stupid my enemies are; Look at how smart my allies are for pointing out how stupid my enemies are; Hello total stranger! You’re an idiot.” PSA from Waco.

• Related: “It means nothing to me. I have no opinion about it, and I don’t care.” Picasso on the moon landing.

• “…probably more seduced by the ambrosia-like scent of maple syrup lingering in the room, but yeah.”

Relative to your interests

• Three Dumb Studies for your consideration - Delightful ideas for scientific studies: “experimenter’s urge: the desire to, quite literally, fuck around and find out.”

• Google’s New AI Tools Are Crushing News Sites - Even the mighty brands are falling: ‘At the New York Times, the share of traffic coming from organic search to the paper’s desktop and mobile websites slid to 36.5% in April 2025 from almost 44% three years earlier, according to Similarweb. The Wall Street Journal’s traffic from organic search was up in April compared with three years prior, Similarweb data show, though as a share of overall traffic it declined to 24% from 29%.’

• 1989: Japan Buys A Used President - This is an incredible piece of writing just one sample: “It is sad and disturbing to witness with what insouciance he now sells the stature which his country afforded him to any country which can afford him. It demeans us all to see him so available, so eager to please for a price, to become ''Rent-a-Ron,' the performing presidential seal.”

• AI in the Enterprise: Priorities, Pressures and Pragmatism - “38% of organizations have deployments under way, and a further 44% plan to follow within the next 12 months. That means more than four in five expect to have generative AI in production by mid-2026, signalling a decisive shift from experimentation to expectation.”

• Patterns Across 5 Years of YC Investing - There’s always money in the Security Stand.

• Bluesky’s decline stems from never hearing from other side - ”When you never hear from the other side, it’s pretty easy to talk yourself into a political dead end. That might be enough for the political dead-enders. But it’s a terrible mistake for any political movement that actually hopes to rack up some durable victories.” // Speaking for myself, I try to “change the channel” and just not pay attention to (follow) anything political. The noise that comes from that is too much and un-actionable. I know: “must be nice…” and all that.

• Starbucks to roll out Microsoft Azure OpenAI assistant for baristas - ”Instead of flipping through manuals or accessing Starbucks’ intranet, baristas will be able to use a tablet behind the counter equipped with Green Dot Assist to get answers to a range of questions, from how to make an iced shaken espresso to troubleshooting equipment errors. Baristas can either type or verbally ask their queries in conversational language.” // Do the baristas need that? After a few weeks, don’t you memorize everything? I guess if there’s a rotating menu, and some weird-o order…

• Did we just make platform engineering much easier by shipping a cloud IDP?Architecture Musings - Google Cloud IDP.

• FinOps Foundation’s FOCUS 1.2 Expands to SaaS, PaaS - Update from FinOps land. Still seems like a lot more people could be doing it.

• The European Sovereign Cloud Day Forecasts Stormy Weather For The Cloud Ecosystem - I watched this conference, as noted in past newsletter episode, the talks were great if this is a topic you’re interested.

• Scaling Vibe-Coding in Enterprise IT: A CTO’s Guide to Navigating Architectural Complexity, Product Management and Governance - There’s some AI middleware considerations (people stuff) towards the end of this how to enterprise vibe coding (in the expansive definition of vibe coding, just using AI to help write code). Part of it is: you still have all the same needs for “day two” and architecting the apps to work in production and be easily updatable (add new features, test it, upgrade the current version [including data], deploy it] for the next version, the next, etc.

• AI’s metrics question - “the real question, as I’ve hinted at a couple of times, is how much LLMs will be used mostly as actual user-facing general-purpose chatbots at all or whether they will mostly be embedded inside other things, in which case trying to measure their use at all will be like trying to measure machine learning, or SQL (how many times a day do you use a database? Who cares?).”

• Related: How is Big AI going to get ROI on that $306 billion?

• Speaking of: “OpenAI has hit $10 billion in annual recurring revenue less than three years after launching its popular ChatGPT chatbot.”

• OpenAI gives ChatGPT access to cloud-based documents and third-party research tools - The way to compete with AI is with the apps, the functionality they have, the outcomes they let you get - the stuff you can do with them.

• How Long Does It Take to Draw a Picture of Every Pub in London? - ”That was true in more ways than one. Ms. Wood, 31, is on a mission to draw every pub in London. She has completed about 300, and has about 2,500 left”

Logoff

I forgot I’d dumped the above as an episode draft. This was just going to be links and wastebook, but here we are!