#software-engineering
50 posts
2026 - October
2 posts2026 - September
2 posts2026 - July
1 post2026 - June
2 posts2026 - May
2 posts2026 - April
1 post2026 - March
2 posts2026 - February
1 post2026 - January
1 post2025 - May
1 post2017 - August
1 post2017 - June
2 posts2017 - May
1 post2015 - December
2 posts2015 - November
1 post2014 - September
1 post2014 - March
1 post2013 - November
1 post2013 - October
1 post2013 - August
1 post2013 - June
1 post2013 - May
2 posts2013 - April
1 post2012 - August
1 post2010 - April
1 post2010 - January
2 posts2009 - November
1 post2009 - September
1 post2009 - July
3 posts2009 - June
1 post2009 - May
1 post2009 - March
1 post2009 - February
1 post2008 - December
2 posts2008 - October
1 post2008 - February
1 post2006 - September
1 post2006 - April
1 post2026 - October
2 postsLLMs + Agents = competent enough
A synthesis of what I've been measuring and writing: LLMs converged. The transformer architecture got near the ceiling, the parameter war is over, and the leap everyone felt in 2025 and 2026 came from the agentic harness (tool calling, prompt caching, long sessions), not from a base intelligence orders of magnitude larger. I retell the timeline from 2019 onward with sources, explain why the agent allows a less dense model, and why that's already competent enough to automate real software engineering.
Understanding the Jev Hype: What Is It For?
Jev is a hosted classifier, and the shock of '193x faster' only exists because so many people use a chat LLM as an expensive classifier. I publish Paulo Câmara's essay (LUA Vision) with his charts, explain what a classifier is and why it's the first thing a computer scientist reaches for, compare the open alternatives Clef and Laya, show the public Decision Index ranking, and close with how to think about this kind of tool in your own project.
2026 - September
2 postsThe Limits of LLMs: Good at Reproducing What Exists. Expensive for What Doesn't
Every LLM was trained only on what's public, but 81.5% of GitHub activity happens in private repositories, and the source code of every proprietary software never made it into training. In my retrocomputing projects I felt firsthand where the machine stalls: with no reference and no oracle, it spins in directed brute force.
Stop Making Excuses and Ship More! With AI, the Premise Changed. Get It.
With frontier LLMs making fewer mistakes than most human devs, the only real excuse left for holding back a deploy is your own review process's bureaucracy. I use my open source projects, cross-platform CI/CD, and real contribution numbers to show how I've been doing this since February.
2026 - July
1 postI Had Fable 5 Analyze the Code of TikTok, Clash of Kings and Gov.br - Understanding Fingerprinting
A static analysis with Fable 5 compares TikTok’s opaque fingerprint, Clash of Kings’ persistent UUID, and gov.br’s privacy precautions, with caveats about the method.
2026 - June
2 postsWhy LLMs Will Fail at Your Company
The deliberately speculative thesis is that LLMs amplify processes where people outsource decisions. The proposed alternative is daily cycles of decision, implementation, testing, and review close to the outcome.
AI Controversy in Open Source Project Contributions - My Take
AI in open source contributions is here to stay, even with AI slop, regressions, and false bugs. The practical answer is to automate triage and auditing without taking the final decision away from humans.
2026 - May
2 postsOpen Source Best Practices with LLMs - The Bare Minimum
Open source projects built with LLMs are only ready with simple installation, reliable tests and CI, and documentation focused on the problem. Standardizing releases and deployments makes automation predictable.
Wrapping Up My AI Marathon: Success or Failure?
After more than 500 hours across 24 repositories, Akita ranks his experiments and concludes Codex and Claude Code support daily coding, but without review and XP, productivity turns into slop.
2026 - April
1 postClean Code for AI Agents
Clean Code carries different weight when an agent is the primary reader: small code, greppable names, provenance context, headless tests, and explicit rules reduce navigation, cost, and errors.
2026 - March
2 postsClaude Code's Source Code Leaked. Here's What We Found Inside.
I analyzed Claude Code’s leaked source map and found hidden features, layered memory, multi-agents, and DRM based on xxHash64. The code also exposed a product that’s difficult to maintain.
Software Is Never 'Done' — 4 Projects, Life After Deploy, and Why One-Shot Prompting Is a Myth
125 post-production commits across 4 projects in 10 days. Real bugs, real users, real iteration. Why one-shot prompting is a myth and software is never 'done'.
2026 - February
1 postIntegration Tests in a Monorepo | Behind The M.Akita Chronicles
How a monorepo running three apps that share a filesystem uses a dedicated integration environment, DevCache, rsynced production data, and preflight checks to catch the bugs unit tests never see.
2026 - January
1 postVibe Code: I Built a Little App 100% with GLM 4.7 (TV Clipboard)
A full walkthrough of vibe coding a tiny Go WebSocket app with GLM 4.7, showing in practice what the 80/20 rule really costs in time, tokens and intervention.
2025 - May
1 postRANT - LLMs are LOOT BOXES!
Why LLMs for real coding behave like gacha loot boxes, and why every incentive in the industry pushes you to burn more tokens.
2017 - August
1 postWhy it is just lazy to bad-mouth Ruby on Rails
Ruby on Rails is neither dead nor the market leader anymore. Its legacy helped define the web's status quo, from Git and continuous testing to the cloud, continuous deployment, and metrics.
2017 - June
2 postsEstimates are Promises - A Better Metaphor
Estimates can become promises when risk is managed: lock in time and cost, prioritize the first 20% of the scope, and use staging deliveries and velocity as gauges for adjusting course.
The Economics of Software Development
For small teams, software economics starts with the smallest product that works, accounting for time while avoiding technical debt, premature microservices, and unnecessary performance optimizations.
2017 - May
1 postProgrammers Guild - Religion and Sports
Technology communities resemble religions and sports teams, with identity and confirmation bias distorting technical choices. Professionals accept trade-offs and take ownership of their decisions.
2015 - December
2 postsEx Manga Downloadr - Part 5: Making it more robust!
MangaFox timeouts expose a flaw in Ex Manga Downloadr’s Workflow. The author uses Task.Supervisor and limited retries to catch HTTPotion errors, leaving GenServer refactoring as technical debt.
Ex Manga Downloadr - Part 4: Learning through Refactoring
Refactoring Ex Manga Downloadr replaces six repeated blocks with macros, improves Floki usage, and uses online tests to validate parsers. The project reaches version 1.0.0 with two fewer files.
2015 - November
1 postObserving Processes in Elixir - The Little Elixir & OTP Guidebook
Using GenServer, Supervisor, and Observer, I show a worker being restarted, a pool of five processes being replenished, and the runtime staying active, though the worker’s state is lost.
2014 - September
1 post[Off-Topic] Agile: the Truth Behind the Method
I argue that Scrum, XP, and other Agile practices cannot save bad teams. They expose risks and problems quickly. Agility depends on committed people, continuous practice, and concrete behavior changes.
2014 - March
1 post[Off-Topic] Lean Is Dead, Long Live Efficiency
Agile and Lean became an industry of courses, certifications, and tools that obscures their principles. The text argues for returning to basics: deliver value, avoid waste, learn, and question.
2013 - November
1 post[Off-Topic] Agile Done Wrong
The text argues that Agile cannot save teams without technical ability and commitment. Planning, pair programming, quality, and responsibility become a performance, and bad professionals need replacing.
2013 - October
1 post[Off-Topic] #noEstimates Debunked
The author separates projects from ongoing operations to defend estimates, objectives, and constraints in projects. To him, constraints drive innovation, but competent execution matters more than any methodology.
2013 - August
1 post[Off-Topic] Estimates Are Promises. Promises Must Be Kept.
The author argues that software estimates are promises, not predictions, and that keeping them requires responsibility for communication, negotiation, time, and obstacles, as well as code.
2013 - June
1 postIntroduction to Agility
Robert Martin recounts how XP, Scrum, and Crystal converged at the 2001 Snowbird meeting into the Agile Manifesto. Akita warns against dogmatic process and consultancies that put method ahead of people and customer value.
2013 - May
2 postsProcesses and Methodologies Won't Help You
Processes and methodologies don’t create good developers. The text argues for continuous practice and competent professionals first, then techniques and processes that can extend their work.
[Translation] Standards: excellence vs. mediocrity
Reporting on an exercise at Toyota, Jason Yip shows that the standard was the champion’s time: four seconds. The text contrasts standards that limit with those that drive excellence.
2013 - April
1 post[Translation] Estimation — The Best We Can Do
Ron Jeffries argues for estimates as ranges of cost and velocity, used to test risks, make decisions every two weeks, and guide the product without becoming a promise or negotiation.
2012 - August
1 post[Off-Topic] The Myth of the "Legacy"
The author criticizes automatically rewriting legacy systems and tells how he fixed an ASP with DCOM in two weeks, preserving the code and winning the client a new project.
2010 - April
1 post[Off-Topic] The Humble Programmer, by Edsger W. Dijkstra
Dijkstra argues for humility before the difficulty of programming: avoid bugs from the start, use abstractions and modest languages, because no tool eliminates complexity.
2010 - January
2 posts[Off-Topic] Reading the Agile Principles
The 12 Agile principles need to be read together: delivering value quickly does not excuse quality, simplicity, collaboration, or continuous improvement. The Manifesto is not a ready-made recipe.
[Off Topic] Dunbar and Cross Functional Teams
In Dunbar's experiments, a multidisciplinary lab solved in 10 minutes a problem that E. coli specialists had spent weeks trying to overcome. Debate and metaphors make new ideas emerge.
2009 - November
1 post[Off-Topic] Restaurants and Technology
The restaurant analogy separates companies whose core business is technology from those that merely use it for support. In the former, programmers should experiment and create tools, as at GitHub.
2009 - September
1 post[Off-Topic] Thinking Makes You Better
Pair programming needs an active pilot and co-pilot, Agility is accountability, and the Spolsky versus Bob Martin spat becomes a case for testing Agile practices, grasping their reasons, and refusing dogma.
2009 - July
3 posts[Translation] What Makes a Good Programmer?
Two translations argue that good programmers prioritize usability, responsibility, collaboration, and delivery. The second adds creative laziness and the humility to ask simple questions while debugging.
[Off-Topic] Intervene or Not? The "Catch-22" of Co-Pilots
Drawing on airplane accidents, the author adapts the PACE protocol for software teams: ask, alert, challenge, and intervene when a manager ignores serious risks.
[Off-Topic] Authority vs. Responsibility
The author argues that separating authority from responsibility dooms processes to failure. Teams accountable for results should decide how they work, with specialists serving as consultants.
2009 - June
1 post[Translation] Advice for Software Development Managers
Gerald Weinberg shows how blame culture and incongruent communication sabotage projects, while congruence improves problem-solving. Change happens through six stages.
2009 - May
1 post[Off-Topic] The 5 Team Dysfunctions in Code
Five team dysfunctions appear in anti-patterns such as null checks, duplication, untested code, and hard-to-understand solutions. Isolated techniques cannot fix a command-and-control organization.
2009 - March
1 post[Off-Topic] Net Negative Producing Programmer
Fabio introduces NNPP, the programmer who creates losses by accumulating technical debt or becoming a single point of failure, and distinguishes this profile from juniors willing to learn.
2009 - February
1 postTranslation: Flaccid Scrum
A translation of Martin Fowler's 'Flaccid Scrum': adopting Scrum without technical practices sinks the team into a messy codebase. Notes on testing, refactoring, frequent integration, and collective ownership.
2008 - December
2 postsTranslation: Technical Debt
Translation of Steve McConnell's classic article on technical debt: accidental vs. strategic debt, the interest that shortcuts charge, and why every debt belongs in the backlog.
Off-Topic: Scientific Method vs Cargo Cult
Cargo culting is repeating structures without understanding why. The fix is testing hypotheses with throwaway prototypes. At YellowPages, four months of prep kept four months of coding from becoming twenty.
2008 - October
1 postOff-Topic: The Agile Manifesto, or How to Become Google
Agility comes from philosophy, trust, and self-organizing teams; methodology alone won't do. From the Agile Manifesto to Conway, Pareto, Wikipedia, and Google: why open source culture breeds innovation.
2008 - February
1 postTranslation: I Don't Know What I Want, But I Know How to Get It
The translation separates iterative development, which expects a solution to be validated and changed, from incremental development, which adds functionality. The point is to plan room to discover, refine, and discard.
2006 - September
1 postEvolution Through Competition
The author argues that criticism and competition drive technology forward. In Rails, limitations around internationalization, legacy systems, and asynchronous tasks have already encouraged community solutions.
2006 - April
1 postRails Manifesto
The author presents Rails as an agile alternative to J2EE’s complexity, built on DRY and Convention over Configuration. The idea is to make Web applications simple and easy to adapt.