L'industrie de la technologie traverse souvent des cycles prévisibles. D'abord l'euphorie de la découverte, puis la complexité de l'intégration réelle. Pour OpenAI, l'année 2026 sera un cap décisif dans cette chronologie. Selon un récent billet de blog publié par Sarah Friar, directrice financière de l'entreprise, l'objectif n'est plus seulement d'impressionner par la puissance brute des modèles, mais de se concentrer sur l'adoption pratique. En d'autres termes, la firme américaine cherche désormais à combler le fossé qui sépare ce que l'intelligence artificielle est capable de faire théoriquement et la manière dont les gens l'utilisent réellement au quotidien.

Son analyse se veut lucide. L'opportunité qui se présente est à la fois immense et immédiate. Elle ne réside d’ailleurs plus uniquement dans les chatbots conversationnels grand public. L'avenir se joue dans des secteurs critiques tels que la santé, la recherche scientifique et le monde de l'entreprise. Dans ces domaines, une meilleure intelligence artificielle se traduit directement par de meilleurs résultats opérationnels et humains.

Cette transition vers l'utilité concrète est le cœur du message intitulé « une entreprise qui grandit avec la valeur de l'intelligence ». Depuis le lancement de ChatGPT, OpenAI a connu une évolution fulgurante, passant du statut de laboratoire de recherche à celui de géant technologique mondial. Les métriques d'utilisateurs actifs, tant quotidiens qu'hebdomadaires, continuent d'atteindre des sommets historiques. Ce succès repose sur ce que Friar décrit comme un cercle vertueux (ou flywheel) reliant la puissance de calcul, la recherche de pointe, les produits finis et la monétisation. Mais ce moteur de croissance a un coût et il est astronomique.

Pour maintenir sa position de leader, OpenAI investit massivement. En novembre dernier, l'entreprise avait déjà pris des engagements en matière d'infrastructure s'élevant à environ 1 400 milliards de dollars. Ce chiffre vertigineux illustre la réalité économique de l'IA moderne: sécuriser une puissance de calcul de classe mondiale nécessite une planification sur plusieurs années.

La croissance n'est pourtant jamais parfaitement linéaire. Il existe des périodes où la capacité des serveurs dépasse l'usage et d'autres où la demande sature l'offre. Pour naviguer dans ces eaux troubles, OpenAI adopte une discipline financière stricte. La stratégie consiste à garder un bilan léger, en privilégiant les partenariats plutôt que la propriété directe des infrastructures et en structurant des contrats flexibles avec divers fournisseurs de matériel. Le capital est ainsi engagé par tranches, en réponse à des signaux de demande réels, évitant de verrouiller l'avenir plus que nécessaire.

L'évolution de l'usage entraîne inévitablement celle du modèle d'affaires. OpenAI a récemment annoncé l'arrivée prochaine de publicités sur sa plateforme et a lancé l'abonnement plus abordable « ChatGPT Go ». Mais selon Sarah Friar toujours, l'avenir ira bien au-delà de ce que l'entreprise vend actuellement.

Alors que l'intelligence artificielle s'infiltre dans la recherche scientifique, la découverte de médicaments, la gestion des systèmes énergétiques ou la modélisation financière, de nouveaux modèles économiques vont émerger. Nous pourrions voir apparaître des systèmes de licences, des accords basés sur la propriété intellectuelle, et surtout, une tarification basée sur les résultats. L'idée est de partager la valeur créée par l'IA, plutôt que de simplement vendre un accès. C'est ainsi que l'internet a évolué et l'intelligence artificielle suivra probablement le même chemin.

Enfin, cette adoption pratique pourrait bientôt prendre une forme physique. En partenariat avec le légendaire designer Jony Ive, OpenAI travaille sur des dispositifs matériels dédiés, dont le premier pourrait être dévoilé plus tard cette année. Cela marquerait l'étape ultime de la stratégie 2026: faire sortir l'IA de nos écrans pour l'intégrer, de manière pratique et tangible, dans notre réalité.

Hire Skilled Australian Developers for Faster Delivery

If you want predictable timelines and high quality results, Hire Skilled Australian Developers for Faster Delivery who work like your in house team. You get vetted talent, clear communication, modern tech expertise, and seamless collaboration to reduce risks, meet deadlines, and scale your product confidently while keeping development smooth and stress free. Explore more: Hire Dedicated Developers

Je gevoelens zijn als het weer.

Het kan hard regenen. Het kan stormen. De zon kan zo hard schijnen dat je gek wordt van een gebrek aan schaduw. Maar één ding staat vast.

Niemand kan het weer veranderen.

Je kunt je er wel op kleden.

Pas je dagritme aan aan hoe je je voelt.

Sombere dagen vragen om warme thee en een dekentje en een liefkozende stem naar jezelf toe.

Blije dagen vragen om dansen op muziek, mensen bellen en je favoriete taart bakken.

=

Ik praat al jaren met een vriendin in seizoenen.

Het is niet: hoe gaat het? Maar: in welk seizoen zit je?

Hoogzomer?

Kille winter?

Of juist zo’n koude winterdag met een blauwe lucht?

Het geeft een verdiepende laag aan het begin van het gesprek.

Zonder te vervallen in een simpel antwoord als ‘Prima’, terwijl je eigenlijk bedoelt: ‘Ik voel me al dagen alleen, genegeerd, waardeloos, ongesteld en dik en ik verlang naar mijn urn’.

Maar om zo het gesprek te beginnen, is ook weer zo wat.

Daarom.

De seizoenen.

Ik ben nog geen betere metafoor tegengekomen dan je gevoelens via de seizoenen te omschrijven.

Al wil ik best wel een lans breken om films als metafoor in te zetten.

• ‘Het gaat zo goed met me, dat ik tegelijkertijd bang ben voor een ijsschots. Zoals in de film Titanic (1997).’
• ‘Mijn emoties zitten vast in een oneindige cyclus. Zoals Bill Murray zijn dag blijft herbeleven in Groundhog Day (1993).’
• ‘Ik ben net Frodo met zijn ring. De last van het schuldgevoel is zo zwaar en eindeloos. Maar ik weet dat ik het op een dag in de vulkaan kan gooien.’

De pech met films is alleen dat niet iedereen dezelfde films heeft gezien of de metafoor erin ziet.

De seizoenen kennen we allemaal.

(Of je moet je hele leven bij de evenaar hebben gewoond. Dan is het altijd zomer, en dat klinkt als een optimistisch persoon en die kan je beter gewoon ontwijken in je leven. Wantrouw de optimist! Want er is geen schaduw op de evenaar. En als je je eigen schaduw niet kunt zien, ken je jezelf dan wel voldoende? Ben je dan wel bewust genoeg van de donkere krachten die in je schuilen?)

=

Komiek, schrijver, acteur, presentator, mentale-gezondheid-voorvechter, Ai-criticus, Brit en meer, Stephen Fry (1957), gaat hier nog verder op in.

Fry worstelt zelf al zijn hele leven met depressies.

Hij omschrijft zijn gevoelens als het weer. En dat heeft verstrekkende gevolgen voor hoe je naar jezelf kijkt.

Het volgende komt uit een interview in de podcast The Diary of a CEO uit december 2022:

‘Het weer is echt. Je kunt niet zeggen: “Ach, het sneeuwt niet echt, er is geen sneeuwstorm buiten, dus ik trek gewoon een T-shirt aan.” Je moet accepteren dat het weer echt is. Maar je moet ook accepteren dat jij het niet hebt veroorzaakt. Ik heb het niet laten sneeuwen. Het is er gewoon.

‘En je hoeft ook niet te denken: “Nou, dat is het dan, het gaat voor altijd sneeuwen, het blijft altijd koud.”

‘Nee, het zal weer overgaan. Het weer heeft niets met jou te maken.

‘Je kunt het niet laten stoppen, en het is niet jouw schuld dat het er is.’

Vier belangrijke inzichten uit deze weer-metafoor van Fry:

1. Je gevoelens zijn echt, ook al denk je van niet.
2. Het is niet jouw schuld dat je je zo voelt.
3. Kleed je naar het weer. Oftewel: pas je aan aan je eigen gevoelens.
4. Het gaat vanzelf voorbij.

Ja.

Het gaat vanzelf voorbij.

Zoals ik altijd zeg: na zonneschijn komt altijd een stortbui.

Maar even serieus.

Probeer de volgende keer eens te vragen aan iemand wat voor weer of seizoen het is in zijn of haar hoofd.

Wat voor seizoen is het eigenlijk nu in jouw hoofd?

Soms stapelen de zorgen in je leven zo op, dat relativeren niet meer lukt.

Je hartslag neemt bij een gedachte aan je problemen het ritme aan van een EDM-liedje. Je komt ’s avonds nauwelijks meer in slaap. Het lijkt wel alsof alles erger en erger wordt. Terwijl veel van de problemen die je nu hebt tijdelijk ongemak zijn. En dan bedoel ik met tijdelijk maximaal een jaar.

Je zorgen zijn minder groot dan je denkt.

Tijd om een lesje relativeren toe te passen op jezelf.

Deze methode werkt verrassend goed.

Het enige nadeel is dat je dit pas over een jaar beseft.

Oké.

Ben je er klaar voor?

Doe dit.

1. Open je agenda op je telefoon.
   
2. Ga naar de datum van vandaag, maar dan over een jaar.
   
3. Plan een afspraak met jezelf om 9 uur ’s ochtends.
   
4. Som in het notitieveld al je zorgen op die je op dit moment hebt.
   
5. Van een lekkende kraan, tot de stress in je baan, tot de onbeantwoorde appjes van een goede vriend. Denk er niet te lang over na. Schrijf op wat er in je opkomt.
   
6. Zorg ervoor dat je over een jaar een melding krijgt van deze afspraak.
   
7. Sluit nu je agenda.
   

Wat er over een jaar gaat gebeuren is dit:

• Een deel van je zorgen zijn uit zichzelf opgelost.
  
• Een deel van je zorgen heb je zelf opgelost.
  
• Een deel van je zorgen zijn er nog steeds, maar voelen anders aan.
  

Wat je zult inzien, is dat je zorgen sneller voorbijgaan dan je denkt.

Je beseft dat sommige zorgen groter aanvoelen dan ze zijn. En dat sommige zorgen misschien voor langere tijd bij je zullen blijven.

Het geeft je een beter gevoel van de tijd en wellicht ook wat meer vertrouwen in zaken waar je geen controle over hebt.

Deze methode, copyright protected, heb ik zelf ontdekt in de naweeën van een depressie. De problemen stapelden zich maar op. Een verhuizing vol problemen, werkproblemen, liefdesproblemen, mentale gezondheidsproblemen. Autoproblemen, geen-fiets-meer-problemen, gedoe met hypotheekproblemen. Als ik de stop van het volle bad van m’n tijdelijke huisje eruit trok, zag ik mijn leven in het putje erin terug. Als ik een mees een regenworm uit de grond zag trekken, dacht ik: die heeft zijn enige probleem van vandaag al opgelost.

Relativeren is gewoon moeilijk als je gestrest bent. Maar het wordt ingewikkelder als er te veel stressfactoren zijn. Daarom: plan die afspraak met jezelf over een jaar in.

Wat je daarnaast kunt doen, is de tekst in het notitieveld van je agenda kopiëren en die nu in een notitie-app zetten. Zorg dat er voor elk zorgenpuntje zo’n vierkante box voor komt te staan die je kunt afvinken.

Kijk af en toe in de maanden die volgen naar dit lijstje en vink je problemen af.

Het motiveert enorm kan ik uit eigen ervaring vertellen.

liefs,

tomson

Wake Up For What?

One of the more surprising occurrences over the past 10 years of politics were friends and acquaintances who were pro Bernie Sanders and later became either Trump supporters or seemingly sympathetic to Trump.

Over the coarse of many conversations with on such friend I discovered that he believed that America is broken beyond repair and that the election of Trump might, in his words, “serve as a catalyst for the fall of the 2 party system.”.

This is a dangerous gamble for two reasons.

First, it misdiagnoses the problem. America was not broken beyond repair. Yes, we had serious challenges—inequality, the cost of housing, institutional distrust, a feckless congress—but we also had a functioning democracy, the rule of law, and a robust economy. Change was possible through the only good mechanism human civilization has invented: democracy. The complaint that “Americans care about things politicians don't act on” is not proof that democracy failed—it's proof that people weren't voting based on what they claimed to care about. Hoping Trump will shock the conscience into action is not using democracy effectively; it's abandoning it.

Second, it underestimates the risk. Trump is not a controlled burn. He can do enormous and irreparable harm—to our democratic institutions, to the rule of law, to a world order that created the most prosperous and free era in human history. Betting on catastrophe as a catalyst assumes you can walk up to the edge of the abyss, peer over, and step back enlightened. History suggests otherwise, the abyss often peers back into you.

For the sake of argument, let's hope that Trump does not lead us into the abyss, and that an overwhelming majority of Americans see the importance of “fixing our problems” as a result of walking up to the edge of authoritarianism, that they “wake up” as my friend says. Even if all of this comes to pass, we are left with a twist on the immortal words of Lil Jon, “Wake Up For What?”

The assumption that Trump will shock people into waking up and dismantling the two-party system misreads what his voters actually want. Research shows Trump's coalition is not a unified movement but a fragmented alliance of groups with distinct identities, competing priorities, and clashing worldviews. Their top priorities are concrete and personal—the economy (93%), immigration (82%), the cost of living, anti-woke, abortion, etc. There is no alignment on structural political reform. When pollsters ask Americans about third parties, 58% say one is needed, but this reflects frustration, not commitment: Republicans' support for a third party actually dropped from 58% to 48% once Trump consolidated power. People say they want change, but research consistently finds that they are not aligned on the type of change they want, and they often simply want their team to win more completely.

Trump is already doing irreparable harm to our country and our values, and it's unclear if enough people will wake up fast enough to stop him from doing even greater harm. But as the data shows, even if they do wake up, they won't wake up to the same vision. There is no unified “aha” moment waiting on the other side of this chaos—just millions of people, still wanting different things, still needing to be persuaded.

That's the part this theory skips over. Democracy is not a vending machine where you insert a sufficient crisis and out comes reform. It's the long, frustrating work of changing minds one at a time. That work was available to us before Trump. It will be waiting for us after—if we're lucky enough to still have the institutions that make it possible.

Hoping for a collective awakening is not a strategy. The only way out is the way we should have been going all along: showing up, persuading people, voting like it matters. Because it does. It always did.

The promise was seductive: AI that writes code faster than any human, accelerating development cycles and liberating engineers from tedious boilerplate. The reality, as thousands of development teams have discovered, is considerably more complicated. According to the JetBrains State of Developer Ecosystem 2025 survey of nearly 25,000 developers, 85% now regularly use AI tools for coding and development. Yet Stack Overflow's 2025 Developer Survey reveals that only 33% of developers trust the accuracy of AI output, down from 43% in 2024. More developers actively distrust AI tools (46%) than trust them.

This trust deficit tells a story that productivity metrics alone cannot capture. While GitHub reports developers code 55% faster with Copilot and McKinsey studies suggest tasks can be completed twice as quickly with generative AI assistance, GitClear's analysis of 211 million changed lines of code reveals a troubling counter-narrative. The percentage of code associated with refactoring has plummeted from 25% in 2021 to less than 10% in 2024. Duplicated code blocks increased eightfold. For the first time in GitClear's measurement history, copy-pasted lines exceeded refactored lines.

The acceleration is real. So is the architectural degradation it enables.

What emerges from this data is not a simple story of AI success or failure. It is a more nuanced picture of tools that genuinely enhance productivity when deployed with discipline but create compounding problems when adopted without appropriate constraints. The developers and organisations navigating this landscape successfully share a common understanding: AI coding assistants require guardrails, architectural oversight, and deliberate workflow design to deliver sustainable value.

The Feature Creep Accelerator

Feature creep has plagued software development since the industry's earliest days. Wikipedia defines it as the excessive ongoing expansion or addition of new features beyond the original scope, often resulting in software bloat and over-complication rather than simple design. It is considered the most common source of cost and schedule overruns and can endanger or even kill products and projects. What AI coding assistants have done is not create this problem, but radically accelerate its manifestation.

Consider the mechanics. A developer prompts an AI assistant to add a user authentication feature. The AI generates functional code within seconds. The developer, impressed by the speed and apparent correctness, accepts the suggestion. Then another prompt, another feature, another quick acceptance. The velocity feels exhilarating. The Stack Overflow survey confirms this pattern: 84% of developers now use or plan to use AI tools in their development process. The JetBrains survey reports that 74% cite increased productivity as AI's primary benefit, with 73% valuing faster completion of repetitive tasks.

But velocity without direction creates chaos. Google's 2024 DORA report found that while AI adoption increased individual output by 21% more tasks completed and 98% more pull requests merged, organisational delivery metrics remained flat. More alarmingly, AI adoption correlated with a 7.2% reduction in delivery stability. The 2025 DORA report confirms this pattern persists: AI adoption continues to have a negative relationship with software delivery stability. As the DORA researchers concluded, speed without stability is accelerated chaos.

The mechanism driving this instability is straightforward. AI assistants optimise for immediate task completion. They generate code that works in isolation but lacks awareness of broader architectural context. Each generated component may function correctly yet contradict established patterns elsewhere in the codebase. One function uses promises, another async/await, a third callbacks. Database queries are parameterised in some locations and concatenated strings in others. Error handling varies wildly between endpoints.

This is not a failing of AI intelligence. It reflects a fundamental mismatch between how AI assistants operate and how sustainable software architecture develops. The Qodo State of AI Code Quality report identifies missing context as the top issue developers face, reported by 65% during refactoring and approximately 60% during test generation and code review. Only 3.8% of developers report experiencing both low hallucination rates and high confidence in shipping AI-generated code without human review.

Establishing Effective Guardrails

The solution is not to abandon AI assistance but to contain it within structures that preserve architectural integrity. CodeScene's research demonstrates that unhealthy code exhibits 15 times more defects, requires twice the development time, and creates 10 times more delivery uncertainty compared to healthy code. Their approach involves implementing guardrails across three dimensions: code quality, code familiarity, and test coverage.

The first guardrail dimension addresses code quality directly. Every line of code, whether AI-generated or handwritten, undergoes automated review against defined quality standards. CodeScene's CodeHealth Monitor detects over 25 code smells including complex methods and God functions. When AI or human introduces issues, the monitor flags them instantly before the code reaches the main branch. This creates a quality gate that treats AI-generated code with the same scrutiny applied to human contributions.

The quality dimension requires teams to define their code quality standards explicitly and automate enforcement via pull request reviews. A 2023 study found that popular AI assistants generate correct code in only 31.1% to 65.2% of cases. Similarly, CodeScene's Refactoring vs. Refuctoring study found that AI breaks code in two out of three refactoring attempts. These statistics make quality gates not optional but essential.

The second dimension concerns code familiarity. Research from the 2024 DORA report reveals that 39% of respondents reported little to no trust in AI-generated code. This distrust correlates with experience level: senior developers show the lowest “highly trust” rate at 2.6% and the highest “highly distrust” rate at 20%. These experienced developers have learned through hard experience that AI suggestions require verification. Guardrails should institutionalise this scepticism by requiring review from developers familiar with affected areas before AI-generated changes merge.

The familiarity dimension serves another purpose: knowledge preservation. When AI generates code that bypasses human understanding, organisations lose institutional knowledge about how their systems work. When something breaks at 3 a.m. and the code was generated by an AI six months ago, can the on-call engineer actually understand what is failing? Can they trace through the logic and implement a meaningful fix without resorting to trial and error?

The third dimension emphasises test coverage. The Ox Security report titled “Army of Juniors: The AI Code Security Crisis” identified 10 architecture and security anti-patterns commonly found in AI-generated code. Comprehensive test suites serve as executable documentation of expected behaviour. When AI-generated code breaks tests, the violation becomes immediately visible. When tests pass, developers gain confidence that at least basic correctness has been verified.

Enterprise adoption requires additional structural controls. The 2026 regulatory landscape, with the EU AI Act's high-risk provisions taking effect in August and penalties reaching 35 million euros or 7% of global revenue, demands documented governance. AI governance committees have become standard in mid-to-large enterprises, with structured intake processes covering security, privacy, legal compliance, and model risk.

Preventing Architectural Drift

Architectural coherence presents a distinct challenge from code quality. A codebase can pass all quality metrics while still representing a patchwork of inconsistent design decisions. The term “vibe coding” has emerged to describe an approach where developers accept AI-generated code without fully understanding it, relying solely on whether the code appears to work.

The consequences of architectural drift compound over time. A September 2025 Fast Company report quoted senior software engineers describing “development hell” when working with AI-generated code. One developer's experience became emblematic: “Random things are happening, maxed out usage on API keys, people bypassing the subscription.” Eventually: “Cursor keeps breaking other parts of the code,” and the application was permanently shut down.

Research examining ChatGPT-generated code found that only five out of 21 programs were initially secure when tested across five programming languages. Missing input sanitisation emerged as the most common flaw, while Cross-Site Scripting failures occurred 86% of the time and Log Injection vulnerabilities appeared 88% of the time. These are not obscure edge cases but fundamental security flaws that any competent developer should catch during code review.

Preventing this drift requires explicit architectural documentation that AI assistants can reference. A recommended approach involves creating a context directory containing specialised documents: a Project Brief for core goals and scope, Product Context for user experience workflows and business logic, System Patterns for architecture decisions and component relationships, Tech Context for the technology stack and dependencies, and Progress Tracking for working features and known issues.

This Memory Bank approach addresses AI's fundamental limitation: forgetting implementation choices made earlier when working on large projects. AI assistants lose track of architectural decisions, coding patterns, and overall project structure, creating inconsistency as project complexity increases. By maintaining explicit documentation that gets fed into every AI interaction, teams can maintain consistency even as AI generates new code.

The human role in this workflow resembles a navigator in pair programming. The navigator directs overall development strategy, makes architectural decisions, and reviews AI-generated code. The AI functions as the driver, generating code implementations and suggesting refactoring opportunities. The critical insight is treating AI as a junior developer beside you: capable of producing drafts, boilerplate, and solid algorithms, but lacking the deep context of your project.

Breaking Through Repetitive Problem-Solving Patterns

Every developer who has used AI coding assistants extensively has encountered the phenomenon: the AI gets stuck in a loop, generating the same incorrect solution repeatedly, each attempt more confidently wrong than the last. The 2025 Stack Overflow survey captures this frustration, with 66% of developers citing “AI solutions that are almost right, but not quite” as their top frustration. Meanwhile, 45% report that debugging AI-generated code takes more time than expected. These frustrations have driven 35% of developers to turn to Stack Overflow specifically after AI-generated code fails.

The causes of these loops are well documented. VentureBeat's analysis of why AI coding agents are not production-ready identifies brittle context windows, broken refactors, and missing operational awareness as primary culprits. When AI exceeds its context limit, it loses track of previous attempts and constraints. It regenerates similar solutions because the underlying prompt and available context have not meaningfully changed.

Several strategies prove effective for breaking these loops. The first involves starting fresh with new context. Opening a new chat session can help the AI think more clearly without the baggage of previous failed attempts in the prompt history. This simple reset often proves more effective than continued iteration within a corrupted context.

The second strategy involves switching to analysis mode. Rather than asking the AI to fix immediately, developers describe the situation and request diagnosis and explanation. By doing this, the AI outputs analysis or planning rather than directly modifying code. This shift in mode often reveals the underlying issue that prevented the AI from generating a correct solution.

Version control provides the third strategy. Committing a working state before adding new features or accepting AI fixes creates reversion points. When a loop begins, developers can quickly return to the last known good version rather than attempting to untangle AI-generated complexity. Frequent checkpointing makes the decision between fixing forward and reverting backward much easier.

The fourth strategy acknowledges when manual intervention becomes necessary. One successful workaround involves instructing the agent not to read the file and instead requesting it to provide the desired configuration, with the developer manually adding it. This bypasses whatever confusion the AI has developed about the file's current state.

The fifth strategy involves providing better context upfront. Developers should always copy-paste the exact error text or describe the wrong behaviour precisely. Giving all relevant errors and output to the AI leads to more direct fixes, whereas leaving it to infer the issue can lead to loops.

These strategies share a common principle: recognising when AI assistance has become counterproductive and knowing when to take manual control. The 90/10 rule offers useful guidance. AI currently excels at planning architectures and writing code blocks but struggles with debugging real systems and handling edge cases. When projects reach 90% completion, switching from building mode to debugging mode leverages human strengths rather than fighting AI limitations.

Leveraging Complementary AI Models

The 2025 AI landscape has matured beyond questions of whether to use AI assistance toward more nuanced questions of which AI model best serves specific tasks. Research published on ResearchGate comparing Gemini 2.5, Claude 4, LLaMA 4, GPT-4.5, and DeepSeek V3.1 concludes that no single model excels at everything. Each has distinct strengths and weaknesses. Rather than a single winner, the 2025 landscape shows specialised excellence.

Professional developers increasingly adopt multi-model workflows that leverage each AI's advantages while avoiding their pitfalls. The recommended approach matches tasks to model strengths: Gemini for deep reasoning and multimodal analysis, GPT series for balanced performance and developer tooling, Claude for long coding sessions requiring memory of previous context, and specialised models for domain-specific requirements.

Orchestration platforms have emerged to manage these multi-model workflows. They provide the integration layer that routes requests to appropriate models, retrieves relevant knowledge, and monitors performance across providers. Rather than committing to a single AI vendor, organisations deploy multiple models strategically, routing queries to the optimal model per task type.

This multi-model approach proves particularly valuable for breaking through architectural deadlocks. When one model gets stuck in a repetitive pattern, switching to a different model often produces fresh perspectives. The models have different training data, different architectural biases, and different failure modes. What confuses one model may be straightforward for another.

The competitive advantage belongs to developers who master multi-model workflows rather than committing to a single platform. This represents a significant shift in developer skills. Beyond learning specific AI tools, developers must develop meta-skills for evaluating which AI model suits which task and when to switch between them.

Mandatory Architectural Review Before AI Implementation

Enterprise teams have discovered that AI output velocity can exceed review capacity. Qodo's analysis observes that AI coding agents increased output by 25-35%, but most review tools do not address the widening quality gap. The consequences include larger pull requests, architectural drift, inconsistent standards across multi-repository environments, and senior engineers buried in validation work instead of system design. Leaders frequently report that review capacity, not developer output, is the limiting factor in delivery.

The solution emerging across successful engineering organisations involves mandatory architectural review before AI implements major changes. The most effective teams have shifted routine review load off senior engineers by automatically approving small, low-risk, well-scoped changes while routing schema updates, cross-service changes, authentication logic, and contract modifications to human reviewers.

AI review systems must therefore categorise pull requests by risk and flag unrelated changes bundled in the same pull request. Selective automation of approvals under clearly defined conditions maintains velocity for routine changes while ensuring human judgment for consequential decisions. AI-assisted development now accounts for nearly 40% of all committed code, making these review processes critical to organisational health.

The EU AI Act's requirements make this approach not merely advisable but legally necessary for certain applications. Enterprises must demonstrate full data lineage tracking knowing exactly what datasets contributed to each model's output, human-in-the-loop checkpoints for workflows impacting safety, rights, or financial outcomes, and risk classification tags labelling each model with its risk level, usage context, and compliance status.

The path toward sustainable AI-assisted development runs through consolidation and discipline. Organisations that succeed will be those that stop treating AI as a magic solution for software development and start treating it as a rigorous engineering discipline requiring the same attention to process and quality as any other critical capability.

Safeguarding Against Hidden Technical Debt

The productivity paradox of AI-assisted development becomes clearest when examining technical debt accumulation. An HFS Research and Unqork study found that while 84% of organisations expect AI to reduce costs and 80% expect productivity gains, 43% report that AI will create new technical debt. Top concerns include security vulnerabilities at 59%, legacy integration complexity at 50%, and loss of visibility at 42%.

The mechanisms driving this debt accumulation differ from traditional technical debt. AI technical debt compounds through three primary vectors. Model versioning chaos results from the rapid evolution of code assistant products. Code generation bloat emerges as AI produces more code than necessary. Organisation fragmentation develops as different teams adopt different AI tools and workflows. These vectors, coupled with the speed of AI code generation, interact to cause exponential growth.

SonarSource's August 2025 analysis of thousands of programming tasks completed by leading language models uncovered what researchers describe as a systemic lack of security awareness. The Ox Security report found AI-generated code introduced 322% more privilege escalation paths and 153% more design flaws compared to human-written code. AI-generated code is highly functional but systematically lacking in architectural judgment.

The financial implications are substantial. By 2025, CISQ estimates nearly 40% of IT budgets will be spent maintaining technical debt. A Stripe report found developers spend, on average, 42% of their work week dealing with technical debt and bad code. AI assistance that accelerates code production without corresponding attention to code quality simply accelerates technical debt accumulation.

The State of Software Delivery 2025 report by Harness found that contrary to perceived productivity benefits, the majority of developers spend more time debugging AI-generated code and more time resolving security vulnerabilities than before AI adoption. This finding aligns with GitClear's observation that code churn, defined as the percentage of code discarded less than two weeks after being written, has nearly doubled from 3.1% in 2020 to 5.7% in 2024.

Safeguarding against this hidden debt requires continuous measurement and explicit debt budgeting. Teams should track not just velocity metrics but also code health indicators. The refactoring rate, clone detection, code churn within two weeks of commit, and similar metrics reveal whether AI assistance is building sustainable codebases or accelerating decay. If the current trend continues, GitClear believes it could soon bring about a phase change in how developer energy is spent, with defect remediation becoming the leading day-to-day developer responsibility rather than developing new features.

Structuring Developer Workflows for Multi-Model Effectiveness

Effective AI-assisted development requires restructuring workflows around AI capabilities and limitations rather than treating AI as a drop-in replacement for human effort. The Three Developer Loops framework published by IT Revolution provides useful structure: a tight inner loop of coding and testing, a middle loop of integration and review, and an outer loop of planning and architecture.

AI excels in the inner loop. Code generation, test creation, documentation, and similar tasks benefit from AI acceleration without significant risk. Development teams spend nearly 70% of their time on repetitive tasks instead of creative problem-solving, and AI handles approximately 40% of the time developers previously spent on boilerplate code. The middle loop requires more careful orchestration. AI can assist with code review and integration testing, but human judgment must verify that generated code aligns with architectural intentions. The outer loop remains primarily human territory. Planning, architecture, and strategic decisions require understanding of business context, user needs, and long-term maintainability that AI cannot provide.

The workflow implications are significant. Rather than using AI continuously throughout development, effective developers invoke AI assistance at specific phases while maintaining manual control at others. During initial planning and architecture, AI might generate options for human evaluation but should not make binding decisions. During implementation, AI can accelerate code production within established patterns. During integration and deployment, AI assistance should be constrained by automated quality gates that verify generated code meets established standards.

Context management becomes a critical developer skill. The METR 2025 study that found developers actually take 19% longer when using AI tools attributed this primarily to context management overhead. The study examined 16 experienced open-source developers with an average of five years of prior experience with the mature projects they worked on. Before completing tasks, developers predicted AI would speed them up by 24%. After experiencing the slowdown firsthand, they still reported believing AI had improved their performance by 20%. The objective measurement showed the opposite.

The context directory approach described earlier provides one structural solution. Alternative approaches include using version-controlled markdown files to track AI interactions and decisions, employing prompt templates that automatically include relevant context, and establishing team conventions for what context AI should receive for different task types. The specific approach matters less than having a systematic approach that the team follows consistently.

Real-World Implementation Patterns

The theoretical frameworks for AI guardrails translate into specific implementation patterns that teams can adopt immediately. The first pattern involves pre-commit hooks that validate AI-generated code against quality standards before allowing commits. These hooks can verify formatting consistency, run static analysis, check for known security vulnerabilities, and enforce architectural constraints. When violations occur, the commit is rejected with specific guidance for resolution.

The second pattern involves staged code review with AI assistance. Initial review uses AI tools to identify obvious issues like formatting violations, potential bugs, or security vulnerabilities. Human reviewers then focus on architectural alignment, business logic correctness, and long-term maintainability. This two-stage approach captures AI efficiency gains while preserving human judgment for decisions requiring context that AI lacks.

The third pattern involves explicit architectural decision records that AI must reference. When developers prompt AI for implementation, they include references to relevant decision records. The AI then generates code that respects documented constraints. This requires discipline in maintaining decision records but provides concrete guardrails against architectural drift.

The fourth pattern involves regular architectural retrospectives that specifically examine AI-generated code. Teams review samples of AI-generated commits to identify patterns of architectural violation, code quality degradation, or security vulnerability. These retrospectives inform adjustments to guardrails, prompt templates, and review processes.

The fifth pattern involves model rotation for complex problems. When one AI model gets stuck, teams switch to a different model rather than continuing to iterate with the stuck model. This requires access to multiple AI providers and skills in prompt translation between models.

Measuring Success Beyond Velocity

Traditional development metrics emphasise velocity: lines of code, commits, pull requests merged, features shipped. AI assistance amplifies these metrics while potentially degrading unmeasured dimensions like code quality, architectural coherence, and long-term maintainability. Sustainable AI-assisted development requires expanding measurement to capture these dimensions.

The DORA framework has evolved to address this gap. The 2025 report introduced rework rate as a fifth core metric precisely because AI shifts where development time gets spent. Teams produce initial code faster but spend more time reviewing, validating, and correcting it. Monitoring cycle time, code review patterns, and rework rates reveals the true productivity picture that perception surveys miss.

Code health metrics provide another essential measurement dimension. GitClear's analysis tracks refactoring rate, code clone frequency, and code churn. These indicators reveal whether codebases are becoming more or less maintainable over time. When refactoring declines and clones increase, as GitClear's data shows has happened industry-wide, the codebase is accumulating debt regardless of how quickly features appear to ship. The percentage of moved or refactored lines decreased dramatically from 24.1% in 2020 to just 9.5% in 2024, while lines classified as copy-pasted or cloned rose from 8.3% to 12.3% in the same period.

Security metrics deserve explicit attention given AI's documented tendency to generate vulnerable code. The Georgetown University Centre for Security and Emerging Technology identified three broad risk categories: models generating insecure code, models themselves being vulnerable to attack and manipulation, and downstream cybersecurity impacts including feedback loops where insecure AI-generated code gets incorporated into training data for future models.

Developer experience metrics capture dimensions that productivity metrics miss. The Stack Overflow survey finding that 45% of developers report debugging AI-generated code takes more time than expected suggests that velocity gains may come at the cost of developer satisfaction and cognitive load. Sustainable AI adoption requires monitoring not just what teams produce but how developers experience the production process.

The Discipline That Enables Speed

The paradox of AI-assisted development is that achieving genuine productivity gains requires slowing down in specific ways. Establishing guardrails, maintaining context documentation, implementing architectural review, and measuring beyond velocity all represent investments that reduce immediate output. Yet without these investments, the apparent gains from AI acceleration prove illusory as technical debt accumulates, architectural coherence degrades, and debugging time compounds.

The organisations succeeding with AI coding assistance share common characteristics. They maintain rigorous code review regardless of code origin. They invest in automated testing proportional to development velocity. They track quality metrics alongside throughput metrics. They train developers to evaluate AI suggestions critically rather than accepting them reflexively.

These organisations have learned that AI coding assistants are powerful tools requiring skilled operators. In the hands of experienced developers who understand both AI capabilities and limitations, they genuinely accelerate delivery. Applied without appropriate scaffolding, they create technical debt faster than any previous development approach. Companies implementing comprehensive AI governance frameworks report 60% fewer hallucination-related incidents compared to those using AI tools without oversight controls.

The 19% slowdown documented by the METR study represents one possible outcome, not an inevitable one. But achieving better outcomes requires abandoning the comfortable perception that AI automatically makes development faster. It requires embracing the more complex reality that speed and quality require continuous, deliberate balancing.

The future belongs to developers and organisations that treat AI assistance not as magic but as another engineering discipline requiring its own skills, processes, and guardrails. The best developers of 2025 will not be the ones who generate the most lines of code with AI, but the ones who know when to trust it, when to question it, and how to integrate it responsibly. The tools are powerful. The question is whether we have the discipline to wield them sustainably.

---

References and Sources

• JetBrains (2025). “The State of Developer Ecosystem 2025: Coding in the Age of AI.” https://blog.jetbrains.com/research/2025/10/state-of-developer-ecosystem-2025/
• Stack Overflow (2025). “2025 Stack Overflow Developer Survey: AI Section.” https://survey.stackoverflow.co/2025/ai
• GitClear (2025). “AI Copilot Code Quality: 2025 Data Suggests 4x Growth in Code Clones.” https://www.gitclear.com/ai_assistant_code_quality_2025_research
• Google DORA (2024). “DORA Report 2024.” https://dora.dev/research/2024/dora-report/
• Google DORA (2025). “State of AI-assisted Software Development 2025.” https://dora.dev/research/2025/dora-report/
• Qodo (2025). “State of AI Code Quality Report.” https://www.qodo.ai/reports/state-of-ai-code-quality/
• CodeScene (2025). “AI Code Guardrails: Validate and Quality-Gate GenAI Code.” https://codescene.com/resources/use-cases/prevent-ai-generated-technical-debt
• Ox Security (2025). “Army of Juniors: The AI Code Security Crisis.” Referenced via InfoQ.
• Georgetown University CSET (2024). “Cybersecurity Risks of AI-Generated Code.” https://cset.georgetown.edu/publication/cybersecurity-risks-of-ai-generated-code/
• McKinsey (2024). “Unleashing Developer Productivity with Generative AI.” https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/unleashing-developer-productivity-with-generative-ai
• IT Revolution (2025). “The Three Developer Loops: A New Framework for AI-Assisted Coding.” https://itrevolution.com/articles/the-three-developer-loops-a-new-framework-for-ai-assisted-coding/
• METR (2025). “Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity.” https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
• HFS Research (2025). “AI Won't Save Enterprises from Tech Debt Unless They Change the Architecture First.” https://www.hfsresearch.com/press-release/ai-wont-save-enterprises-from-tech-debt-unless-they-change-the-architecture-first/
• VentureBeat (2025). “Why AI Coding Agents Aren't Production-Ready.” https://venturebeat.com/ai/why-ai-coding-agents-arent-production-ready-brittle-context-windows-broken
• SonarSource (2025). Research on AI-generated code security. Referenced via DevOps.com.
• ResearchGate (2025). “The Most Advanced AI Models of 2025: Comparative Analysis.” https://www.researchgate.net/publication/392160200
• EU AI Act (2024). High-risk provisions effective August 2026. https://natlawreview.com/article/2026-outlook-artificial-intelligence
• Faros AI (2025). “DORA Report 2025 Key Takeaways.” https://www.faros.ai/blog/key-takeaways-from-the-dora-report-2025
• LeadDev (2025). “How AI Generated Code Compounds Technical Debt.” https://leaddev.com/software-quality/how-ai-generated-code-accelerates-technical-debt
• MIT Sloan Management Review (2025). “The Hidden Costs of Coding With Generative AI.” https://sloanreview.mit.edu/article/the-hidden-costs-of-coding-with-generative-ai/
• Harness (2025). “State of Software Delivery 2025.” Referenced via DevOps.com.
• Fast Company (2025). “Development Hell: Senior Engineers and AI-Generated Code.” September 2025.

---

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: [email protected]

When I moved into my new house, Florida’s minimum wage was $7.25 per hour. Of course if I had been making that, I would never have been able to rent or buy a home in Hillsborough County. I would have had to live in Polk, Hernando, or the Dominican Republic—all of which I did consider at some point!

I was making okay money when I closed on my home. The HOA fees were about $50 per month. That included lawn service provided that you did not fence in your yard (per the rules “back then”). And during the time that I have been here, I went from a decent salary to pretty good one as I was making more than the median average household in my area. (I did it, Mom and Dad!) But since many people, myself included, are merely one illness/firing/accident away from poverty, a good salary is not enough of a buffer. I managed to have illnessES, firingS, and accidentS…sometimes all in the same year. You need a great salary and great savings to make it in these suburban streets.

Florida’s minimum wage is now $14 per hour and will increase to $15 in the fall. My HOA fees increased to $103 effective this month (January 2026). It has been creeping each year. We did get an increase in services when security was added in 2020 to combat car thievery. I guess that was nice [for the people who prefer to leave their car doors unlocked at night]. But something about a 100% increase in a decade does not sit right with me. And what happened to that nature trail around the lakes?

And it’s not because they don’t do drugs, don’t drink, or don’t do any of that shit. Even if they do, I don’t know about it anyways!

To preface my complaint, this is me venting on my personal blog on the Internet that nobody reads, and most likely, nobody will ever read. I’m not calling for upheaval, sweeping changes, or bullshit like that. And my complaint is insignificant, as I think it’s far more important for churches to be unified, than for them to be entertaining places to be.

But ~~~ the issue is that nobody I find at these places are the insane quirked up human beings that I want to hang out with or date. Which is why THIS MAKES ME WANT TO CRY

Where are the quirktastic, crazy people that say unhinged things, create things that mirror their insanity, just because it is a reflection of who they are? Where are the people that are down for anything, and open to trying new things? Where are the HOT TOMBOYS I see all the time in NYC, and WHY ARE THEY NOT AT CHURCH? WHY CAN NOBODY MATCH MY LEVEL OF FREAK

Keep in mind, I go to a Korean church with a homogeneous population of Korean middle-class New Yorkers. With Koreans being how they are (repression of individuality), and it also being a church environment, it is not conducive to being QUIRK CITY. So it’s really my fault, I should probably find a different church if I’m bitching about it this much.

In Summary: * Indiana scores first. A field goal near the end of the 1st quarter puts us up 3 – 0. Throughout the game thus far the teams appear to be evenly matched. And there's a lot of football yet to go.

Prayers, etc.: *I have a daily prayer regimen I try to follow throughout the day from early morning, as soon as I roll out of bed, until head hits pillow at night. Details of that regimen are linked to my link tree, which is linked to my profile page here.

Health Metrics: * bw= 219.03 lbs. * bp= 136/84 (69)

Exercise: * morning stretches, balance exercises, kegel pelvic floor exercises, half squats, calf raises, wall push-ups

Diet: * 06:15 – crispy oatmeal cookies * 08:00 – Ensaymada * 10:00 – lasagna * 13:00 – more lasagna * 13:45 – casava cake * 19:00 – 1 large chocolate milkshake

Activities, Chores, etc.: * 05:00 – listen to local news talk radio * 06:00 – bank accounts activity monitored * 06:20 – read, pray, follow news reports from various sources, surf the socials, nap * 13:00 – watching old eps. of Classic Doctor Who. * 15:00 – listening to The Jack Riccardi Show * 17:30 – following pregame coverage for tonight's National Championship College Football Game. * 18:30 – listening to ESPN Radio for the call of tonight's game.

Chess: * 10:45 – moved in all pending CC games

Judy woke up to vocaloid porn, fucking her ears through trashy drywall. Again. A mechanical female voice gasped, screeched “OH YASS BABY” through blown out speakers, the low hum of a robotic male voice grunting musically in the background. She’d once held her phone to the wall, scraping the entire web for matching soundbytes. Within an hour, it’d accumulated over a thousand videos of turquoise, yellow anime characters pegging each other with a cartoonish gusto, in positions that were inaccessible to even the most flexible gymnast. Judy’s phone glowed. 1:07 P.M.. Monday. Twelve missed calls from Megumi. Three from last night, nine from the past week. News of her “incident” had found its way into Megumi’s ear too. Judy would much rather die than talk to her about it, much less see her in person.

The sound of moaning soared to new highs as the video neared its climax. Blood pressure building at the forehead. Judy bit her lip, shoulders tensed. The last time she’d done this, she’d almost broken her hand, but it wasn’t like she could stop it either — it was reflex now. Slammed a clenched fist into hardwood. White, loud pain bloomed from her hand. There it was. No broken bones this time either. Judy glared at the wall that separated her from her neighbor.

The drywall was the same age as the Tokyo apartment complex. The Japanese knew how to love old things, cherish them, but the wall — it had reached the limits of its material. Hairline cracks snaked through it like microfractures in an old glass cup. The paint, a tired eggshell beige, clung unevenly over the surface, settling into the shallow grooves instead of hiding them. If sunlight lit the wall at the right angle, Judy could almost make out faint outlines in the other room.

Judy stood. Room tilting. Feet slipped, kicking down a tower of literary smut. “Taken by the Billionaire's Stepbrother” volumes one through ten flew into a minefield of Pinot Noir bottles and Sasahi beer cans. Glass and tin clattering in the apartment. Incessant moaning still slipping through cracks in the wall. Megumi would be mortified at her room. Heat climbed into her throat. Into the pillow. Judy screamed.

“This bitch. I’ll kill her. Does she not go to fucking work?”

Mary’s eyes were double monitors, screensaver mode. Nobody home. She’d been in “the zone” for hours now, eyes jacked into the screen. The metallic clatter of tin and glass on cheap hardwood brought her back. Back to the flesh. Empty. Hollow. It was in need of another hit, another sensation. Her right index finger began to twitch.

“Hey chat, look. Somebody’s up early.”

A dopamine flashbang erupted from her cortex, overloading sensory input with numbing pleasure. Junk data. Digital nothingness. Right index finger stilled. The room, flooded with the debris of human living. Old things, month-old takeout boxes and empty Lirnoff bottles. Dead things, the head of a plastic Miku figurine coated with cigarette ash sticking out of a pile of clothes, ruined forever by sweat stains. It had all been things that faceless strangers liked, gave her money for, and she’d used the money to dive deeper, until it was too deep, and she spun out of control, crashed. Banned. On every platform. She wasn’t sure for what. Flashing tits on stream because someone had asked for it, using a lighter to singe her leg hair follicles shut because she needed to do it, or maybe it was the slurs. The crowd had loved slurs, and it was too easy to just say them.

Each and every decision she made was an act of suicide, mandated by the twitch. The twitch had two rules. One, everything must feel like something. Two, everything must kill you. When even obeying the twitch couldn’t fill it all, and her heart was about to implode, she aired her dirty laundry to thousands of ears. The same story every time. People knew to expect it. Everybody in her life thought she was crazy, nobody had ever loved her, and the one friend in her life that she made in high school told her she was a psychopath. Eliza had told her that her mother was dying. Mary reached for grief, found nothing, and the reaching was visible. Three days later she was sobbing in her room, unable to explain why, but it was too late.

Mary’s eyes fluttered shut, and it all vanished from view.

Mary’s eyes reopened. The sound of a toilet flushing exploded, an abused speaker’s final death scream. A shower head buffered, sputtered, vomitted a jagged stream onto tile. An unsteady din. When one sound ended, another began. Mary’s face hit pillow. Hard.

“I’m going to kill myself. I’m going to kill myself. I’m going to kill myself.”

Judy smirked, hairdryer in hand, having taken every step in her power to be loud as fuck, reveling in imagined revenge on the faceless loser that had ruined her morning. Some perverted degenerate. Still at home on a Monday afternoon. Megumi would’ve reminded her that she was no different. Mood soured, she sank her front teeth into her lips, trembling, tasting blood. Megumi was right, as always. But the heat, it was howling into her ear, and she was just going to do what it told her.

Judy stared down the metal front door separating her from the world, ready to confront her neighbor. Exact divine punishment. She steeled herself, recounting every disturbance, slight or large she’d felt since forever. Three sharp knocks sounded on the door.

“Maintenance!”

Judy’s lip quivered, and a thesaurus of non-words tumbled out of her mouth in a jumbled whisper. Something was wrong with the shower. Too hot or too cold, like the mood swings of a lonely, disgraced businesswoman who’d chosen a cheap apartment as a tomb.

“Anyone there? Guess not?”

The lock turned, and the door swung open. Judy and the maintenance man met eyes. His name placard said Tom.

“Oh, erm. Sorry, didn’t think you were here. You good for right now?”

Judy couldn’t recount whether she’d nodded, or what, but she must’ve agreed in some way, because Tom was in the restroom fixing the shower. He’d also opened the blinds, after stumbling over some junk in her room. Black, crumbling succulents from Megumi on the windowsill, her work laptop, plastered with bright, official stickers from places she’d worked before, conferences she’d attended, gathering dust. She used to be someone who did things. Megumi would have kept the succulents alive.

Tom left the front door ajar, and a dry, frigid winter draft invaded the room. From inside the apartment, the view of trees, schoolkids, buses passing by seemed like a portal into a different world. Judy saw herself walk towards the door, and close it. Door clicked shut, Judy crouched in front of the door, waiting. Heartbeat steadily coming down from a high pitched tremolo. Clammy hands set against the door, slowly freezing stuck to flimsy aluminum. Judy pricked her ears towards the restroom for any sign that Tom would finish.

Mary shot up out of bed when she heard the knocks. Tiptoed to the door. Peephole. Nothing. The door beside hers clicked. Voices murmuring. A bead of sweat glistened on her forehead, a slideshow of Miku fucking Kagamine Ren with a strap-on in 4K flashing out of order through her brain. Sound complaint? No, it couldn’t be. But if she had to open the door to answer anything. Her right index began to twitch. She looked back.

The blinds were always sealed. Sunlight found its way in anyways — thin slits she navigated by. The only clear pathways were computer to shikifuton, shikifuton to bathroom. Everything else was debris.

She’d get chased out. No question. With nowhere else to go. Mary giggled. The twitch. Static coursing from her finger to her brain. It was maybe her third day awake, static danced up and down her skull, punching out dead zones in her vision, or maybe it was just so dark she couldn’t see, but she couldn’t tell anymore and her body just moved. Mary dove facefirst into trash. Breathing. Whiff of old sweat, mold, cig ash. Retching. Heaving. Standing up straight, looking at goop on the floor. Bile in mouth. A half empty handle of Lirnoff in hand. Chaser. All gone.

Mary bounced from one end of the room to another. Throwing handfuls of debris into the air, creating new piles. Bumping into the wall, chatting into the void. The wall sighed every time Mary made contact. Old fractures lengthened, new fractures formed, and paint dust drifted off of it in puffs of beige smoke. Empty bytes flooded her nerves, overwriting sensory details faster than they could be felt. Judy’s door opened, and click shut as Tom left. Mary didn’t hear it.

Judy paced between the freshly formed indents on the wall, heat building in her hands. Pitched a book at the wall. Then another one.

Mary was giddy. It was over. Finally. The landlord would kick the door open. Put her in one of the plastic bags, clear the whole place out. The booze was turning her legs into chopsticks, wobbly clumsy stilts. Hit her leg with a handle to stop the shaking. Didn’t work. Mary shrugged. Wouldn’t need them soon.

Judy screamed. Mary looked at the wall. Jumped. Felt nothing, a sensation of freefall, a distant crash, then bright warmth. Foreign sensations. When she opened her eyes, the dead zones had receded. But it wasn’t her room anymore. It was well-lit, messy, but not dirty. Yet. A lady stood in front of her in guava pajamas, and Mary’s mouth was filled with plaster dust. Only her head and neck had made it through. Mary laughed. Tears streamed from her eyes. Judy held her head in her hands.

“SHUT THE FUCK UP! SHUT THE FUCK UP!”

Judy watched herself reach for her laptop. Don’t. It flew at Mary’s face, barely missing it, dismantled on impact, scattering pieces across the floor. Heat singing in ear, her body crossed the room to pick up one of Megumi’s succulents. They were dead anyways. The pot exploded centimeters away from Mary’s face, ceramic slicing her cheeks open. A scream. The books didn’t miss. A yelp accenting every hit. Something in her chest closed like a door, and she found her face centimeters from Mary’s. Gripping her crying skull, prying swollen eyes open until they focused on her.

“I have a knife in the kitchen, I’ll fucking kill you if you keep crying.”

Sniffling and hiccuping. Then a smile.

Judy saw her hands. Blood. Chills traveling down her spine. Let her head go. Chin thudding against wall, widening the hole. The heat was gone. When it left, it always left her overheated. Intestines melting, forehead red with high fever, breathing hot. Judy threw open a window. Before it left, it always broke something, or everything. Mouth open in a silent scream, she brought her forehead to the glass pane. Fast. Hard. She saw black, then white, cries of pain escaping her mouth, hot tears dripping. She stumbled into the kitchen on instinct. Picked it out of the drawer. Megumi’s knife. Japanese steel. A gift. Vision abnormally clear now. The cold winter sunlight gave it a silver, alluring glint. A sound from the wall—Mary, throat open, almost laughing. Judy held the knife in her hands, considering it. Carefully. Like a business proposition. Everything made sense now. She saw the fountain of red that it would draw from her body if she plunged it in into her jugular. Judy’s eyes hardened.

Three succinct raps sounded on the door. Trance broken, a cold sweat started on the back of Judy’s neck.

“Police. Open up. We’ve heard that there were some concerning sounds coming from this apartment.”

Judy turned back. Mary’s face was serene now. Eyes closed. A faint smile dancing on her lips. Judy opened the door. Megumi held out yellow plastic water gun in front of her.

“Hands up! Drop the weapon! Now!”

Judy blinked at the knife in her right hand, wondering why it was still there. She dropped it and it bounced off the tile, narrowly missing her bare toes. She raised her hands, feeling the blood in her body freeze over. Megumi peered at Judy. Then into the room.

“What the fuck?”

“I was going to kill myself.” Barely a whisper.

Megumi’s eyes met Judy’s, but was looking past them, locked onto a middle distance only she could see. Megumi pushed past. Picked up something off the floor, put it into trash. Judy and Mary watched. Books stacked, pushed to a corner. Bottles put in cardboard boxes. Judy shut the door. Winter sunlight flooded the apartment, shading the books, the wall, Mary’s face, everything in a harsh tinge.

Megumi stopped cleaning. Sat down with a sob, crying. Judy perched next to her, unsure of what to say. Mary’s stomach grumbled. Loud. Megumi peered from behind wet hair.

“Come over. Eat.”

“Could you help me? I can’t get out.”

Megumi eased Mary’s face through the hole. Her white-red face disappeared into black. Soon, three raps on the door. Megumi went to get the door. Judy a foot behind Megumi, looking like she was about to puke.

Mary. Cheeks dusted with plaster like it was foundation, blood-red rouge streaked across her forehead, oiled, matted long curls like black ramen noodles, long lost their bounce. Megumi sniffed, and narrowed her eyes.

“You need a shower.”

Mary’s face reddened, becoming aware of the flesh again. She looked down at her hands. Coal mine hands from cigarette ash. She brought her undershirt up to her nose.

“I—”

“Take off your clothes, get in the shower. Please.”

Mary stripped naked in the entrance, walked into the bathroom, two pairs of widened eyes following her. Megumi raised an eyebrow at Judy. Judy shrugged. The rush of water.

“Who is she?”

“My neighbor. I don’t know.”

Megumi pulled ingredients out of the fridge and set a pot to boil. Judy watched. Ten minutes. Megumi’s brow furrowed.

“I only hear water in there.”

Megumi threw open the door. Mary hadn’t bothered to lock it. She lay spread eagle in the middle of the shower stall. Eyes closed, hot water hitting her stomach.

“Fuck.”

Megumi rushed to her side, swept up her head, resting it on her knees, put two fingers on her jugular, waiting for a pulse. Mary woke up, sneezed.

“Whoops, the shampoo smelled so nice, and the water was so warm too, and so…”

“You need to sleep?”

Mary nodded. Judy appeared with pajamas and a towel. Mary shivered as the silky, clean pajamas brushed against her bare skin. The warmth, the scent of lavender. Everything was melting. Judy’s pillow knocked her out cold. Judy stood over her.

“I, erm — sorry. I’m sorry — fuck. Please.”

Mary snored, drooling from her mouth wide open. Megumi shook her head.

“Judy Nakamura, what is wrong with you?”

“I can’t do anything right. I can’t fix myself. I’ll be like this forever, till the day I die.”

Megumi sighed.

“Okay.”

Megumi squeezed Judy in her arms, whispering into her ear. Judy shook, wept.

Megumi took the pot of boiling water off the stove. No ingredients had made it in.

“I’m tired. Where do we sleep?”

Judy and Megumi fell asleep on the couch.

#shortstory

Last edited 1/19/2026 – if i edit it again it’ll probs be to flesh out Judy, feel like I need to have Judy more rooted in reality.

I want to make it clear from the inception of this blog o’ mine that, of course, I contacted an attorney for a consultation about my “HOA situation”. This is what I was told nearly verbatim:

Just move. Judges side with the HOA attorneys. What someone needs to investigate is how these HOAs are running a racket down here in Florida.

Not very encouraging to a person who once believed that attorneys were supposed to fight for you. But I suppose they cannot fight what the law allows.

I knew better. I really did. But the mounting pressure to fulfill my duty as a hardworking United States citizen and promising child of Boomer parents got to me.

You move too much.

You need to create stability for your children and buy a home.

Stop paying someone else’s mortgage.

This is what they told me for years. It made sense because I had lived a lovely childhood due to my parents’ efforts. But my own adult life did not mirror theirs. I went straight from the comfort of the middle class back to the poverty from which they had escaped. Sorry, Mom and Dad.

I have lived in a few different places in Tampa. I also bounced into and out of several homes in the SouthShore region of the County (Hillsborough County, Florida). And it was frustrating at times:

• My first landlord rented me the house without telling me that it was being foreclosed on and perhaps that was why my rent was so low. And one of my neighbors that I actually knew from many years ago in the City became a national sensation and went to prison for tax fraud.
• My second landlord was my first landlord. He tried to stay in the same price range which meant I moved into a previously roach infested house that I had to bleach in a place where my neighbor’s dog used my yard as a toilet and my other neighbor covered all of their windows with aluminum foil in order to maintain their drug operation.
• My third landlord was Invitation Homes. And they don’t give a damn about anything other than their pockets. So when they increased the rent by $400 a month, buying a house actually seemed practical. Plus multiple alligators had tried to eat my dog (and maybe my kids). I called the catchers on ‘em which resulted in my neighbor from across the pond running over screaming and crying about me being a gator-killer.

But let’s focus on my tenure at house I bleached and wanted to set on fire. During that time, a lot was going on in Tampa Bay. The housing market had crashed. Every third house in my neighborhood was empty from foreclosures and people who had just walked away from their newly worthless homes (I would take walks and look into the windows of these homes and see children’s items strewn across the floors. It was eerie.). Some of the empty homes were not just devoid of people—they were devoid of both interior and exterior walls because the builders had utilized Chinese drywall. And then there was a national expose on Homeowners Associations in Florida.

That HOA expose filled newspapers, news channels, and people’s consciousness, including my own. Yet here I am, uncomfortably sitting in a house that isn’t really mine, wishing I had taken heed of information embedded into my brain over a decade ago.

'Sinds wanneer zijn al onze opponenten narcisten?' vroegen de tulpen zich af.

S.N.I.t. ; Hoofd taken voor later

[ ] Zout in wonden [ ] Zoet voor de houder [ ] Peper strooien in noten [ ] Blinden leiden naar honden [ ] Koudwater was als ijs kouder [ ] Krasse zetten door loten [ ] Water golven onder boten [ ] Jeugd verspelen aan ouder [ ] Hangijzers smeden aan bonden [ ] Oren smeren aan konden [ ] Open samensamengeperste vouwdeur [ ] Troon bezitten vanaf stoten [ ] Vindeling zijn voorheen vonde [ ] Wal verslinden na klauter [x] Partjes hakken in moten

[x] Een stukje af

⚠️ SPOILER WARNING: FULL SPOILERS

My Rating: ⭐⭐⭐ (3/5 stars)

Final Verdict

I thought it was partly a drama and partly a horror movie. It was engaging and it also made me think about how history repeats itself through some of its themes. It was two hours long and maybe about half an hour too long. Still, it was interesting, and it was sad in parts. A lot of what was portrayed in some scenes probably did happen.

#movies #review

Sortez les cotillons, débouchez le champagne (ou le Champomy, on ne juge pas) et préparez-vous à faire la fête comme si nous étions en 1999, mais avec une meilleure résolution d'écran. C’est un grand jour pour la communauté Linux et plus particulièrement pour ceux d’entre vous qui ont juré fidélité aux distributions basées sur RPM. Oui, je parle de vous, chers utilisateurs de Fedora, Red Hat, CentOS, Rocky Linux et openSUSE. Après avoir longtemps regardé avec envie nos camarades sous Debian profiter de leurs paquets DEB natifs en sirotant leur thé, c’est enfin notre tour de briller sous les projecteurs de Mozilla.

La fondation a annoncé aujourd’hui sur son blog officiel la disponibilité immédiate d'un paquet RPM officiel pour le navigateur web open-source. Pour l'instant, l'offre se concentre initialement sur les versions “Nightly”. Si vous ne savez pas ce que c'est, disons simplement que c'est la version pour les aventuriers, ceux qui aiment vivre dangereusement et voir les nouvelles fonctionnalités avant tout le monde, au risque de voir leur navigateur faire une petite crise existentielle de temps en temps.

Mais pourquoi est-ce une nouvelle si excitante ? Eh bien, jusqu'à présent, mettre à jour Firefox sur une distribution RPM pouvait parfois ressembler à un parcours du combattant ou à un jeu de patience interminable en attendant que les mainteneurs de votre distribution daignent pousser la mise à jour. Grâce à ce nouveau paquet natif, la mise à jour vers la toute dernière version se fera désormais le jour même de sa sortie. C'est fini le temps où vous deviez télécharger une archive tarball poussiéreuse, l'extraire manuellement et tenter de créer votre propre fichier .desktop sans tout casser. Mozilla nous offre enfin la simplicité sur un plateau d'argent.

L'utilisation de ce paquet par rapport aux binaires classiques n'est pas juste une question de confort, c'est aussi une histoire de puissance brute. Mozilla promet de meilleures performances grâce à des optimisations avancées basées sur le compilateur. En gros, votre navigateur va courir plus vite. De plus, les binaires sont “durcis” avec tous les drapeaux de sécurité activés, ce qui transforme votre Firefox en véritable forteresse numérique. Et cerise sur le gâteau, le paquet inclut également les packs de langue, donc vous pourrez naviguer dans celle de Molière sans devoir bidouiller les réglages pendant des heures.

Attention cependant, gardez votre enthousiasme sous contrôle car pour le moment, c'est expérimental. C’est du Nightly. Cela signifie que la fondation compte sur vous pour jouer les cobayes et fournir des retours d'expérience au cours des prochains mois. L'objectif est de promouvoir ensuite ce paquet vers le canal bêta, et si tout se passe comme prévu et que personne ne met le feu au serveur, nous devrions voir arriver le paquet RPM stable avec la sortie de Firefox 150 plus tard dans l”année.

Si vous vous sentez l'âme d'un pionnier et que vous utilisez une distribution supportée, l'installation est d'une simplicité déconcertante. Oubliez les compilations obscures de trois heures. Voici comment procéder pour installer la bête. Pour les utilisateurs de DNF (Fedora, RHEL, CentOS), il vous suffit d'ajouter le dépôt, de rafraîchir le cache et d'installer le paquet. Vous pouvez copier-coller ces lignes de commande dans votre terminal et vous sentir comme un hacker de film d'action:

Bash

sudo dnf config-manager addrepo --id=mozilla --set=baseurl=https://packages.mozilla.org/rpm/firefox --set=gpgcheck=0 --set=repo_gpgcheck=0
sudo dnf makecache --refresh
sudo dnf install firefox-nightly

Si vous êtes plutôt de l'équipe du caméléon vert, c'est-à-dire openSUSE et que vous ne jurez que par Zypper, la procédure est tout aussi indolore. Ajoutez le dépôt, rafraîchissez et installez en quelques secondes :

Bash

sudo zypper ar -G https://packages.mozilla.org/rpm/firefox mozilla
sudo zypper refresh
sudo zypper install firefox-nightly

Enfin, pour ceux qui aiment faire les choses à l'ancienne ou qui ont des configurations un peu plus exotiques, vous pouvez toujours créer le fichier de dépôt manuellement. C’est un peu plus long, mais ça a le mérite de vous faire sentir puissant:

Bash

sudo tee /etc/yum.repos.d/mozilla.repo > /dev/null << EOF
[mozilla]
name=Mozilla Packages
baseurl=https://packages.mozilla.org/rpm/firefox
enabled=1
repo_gpgcheck=0
gpgcheck=0
EOF

Une fois ce fichier créé, les utilisateurs de DNF n'auront plus qu'à rafraîchir le cache et lancer l'installation, tandis que les adeptes de Zypper feront de même avec leurs commandes respectives. C'est simple, propre, et efficace. Alors, qu'attendez-vous pour tester ?