iiinova — where your digital self keeps evolving

Alexandru Nedelcu @alexn.org · 27/09/2026

Age of AI is bringing with it some of the dumbest takes imaginable. Newer samples: - Generated code can be assembly; - You no longer need unit tests, or DRY, or abstractions. This is the result of “cargo cult programming”, and of doing stuff based on vibes, which precedes AI.

220

Alexandru Nedelcu @alexn.org · 22/09/2026

My article reached HN 😎 alexn.org/blog/2026/0...

Screenshot of a Hacker News page. Top story: “AI Has No Wisdom and Neither Will You” from alexn.org. It has 178 points, posted by dimonomid 1 hour ago, with 223 comments. Above is the orange Hacker News navigation bar with new, past, comments, ask, show, jobs, and submit links.

040

Alexandru Nedelcu @alexn.org · 21/09/2026

My #OpenCode configuration now has a custom plugin that evaluates shell commands for safety via #Jev. It evaluates if shell expressions are within the agent's permissions, according to its configuration & system prompt. github.com/alexandru/...

Code editor showing a JSON configuration file. The snippet defines a "shell-safety" plugin with options like endpoint and model, and configures "Explorer" and "Librarian" agents.

000

Alexandru Nedelcu @alexn.org · 09/09/2026

This is how a setup with delegation to cheaper models looks like, pics show a single session. It was harder to balance back in the days of Haiku 4.5. A dumb model can do damage that the smarter model won't fix, but nowadays, Luna, or DeepSeek are quite capable. #LLM #AI

A table titled "Agents" listing four agents sorted by cost. Orchestrator is the most expensive at $100.04, followed by Explorer at $18.53, Junior at $10.77, and Librarian at $4.06. The table includes token usage and message counts.

A table titled "Models" listing usage stats for two OpenAI models. Row 1: gpt-5.6-sol with 3.3M input, 243K output, 188.6M total, 187ms/1K, and cost $98.88. Row 2: gpt-5.6-luna-fast with 55.7M input, 2.3M output, 885.3M total, 99ms/1K, and cost $33.37.

010

Alexandru Nedelcu @alexn.org · 03/08/2026

#OpenCode Go is a good subscription that I've been recommending, but, with the latest updates for DeepSeek, I'm not sure that they know what their value proposition actually is (DeepSeek's official API is cheaper). That, plus they just broke all my flows. Thanks, but no, thanks.

000

Alexandru Nedelcu @alexn.org · 16/07/2026

I've never seen an engineer "attempt" to override $HOME and then delete it. What a silly thing to do. "The Daily WTF" notwithstanding, meat-space having stories of blunders, I've never worked with engineers that get this silly. But hey, at least this is fast & cheap 😏

Tweet by Tibo saying:

"On file deletions. We’ve investigated a handful of reports where GPT-5.6 unexpectedly deleted files.

What we have found is that this most commonly occurs when:
- Full access mode is enabled and codex is run without sandboxing protections, including without auto review being enabled
- The model attempts to override the $HOME env var to define a temporary directory.
- The model makes an honest mistake and mistakenly deletes $HOME instead."

040

Alexandru Nedelcu @alexn.org · 30/06/2026

In #Java land, Logback, one of the most foundational libraries, removes support for certain configs with a patch version, so upgrading this library (as demanded by scanners like Checkmarx), can break your logs. The SEMVER is strong with this one 🥲

242

Alexandru Nedelcu @alexn.org · 26/06/2026

Flow is a Typescript alternative by Facebook. They've rewritten from OCaml in Rust, w/ a single engineer steering AI. The cost of switching tech stacks becomes cheap, converging to zero and this is one trend that's changing the industry.

3110

Alexandru Nedelcu @alexn.org · 11/06/2026

... as opposed to spending all day looking busy in meetings 😀

Screenshot of the Hacker News website showing the orange header with navigation links and a top story headline: "Workers are spending over 6 hours a week botsitting AI, fueling job frustration" from businessinsider.com, posted 3 hours ago with 209 points and 178 comments.

100

Alexandru Nedelcu @alexn.org · 09/06/2026

Last night I vibe-coded support for threads of posts, supported on #Mastodon, #Bluesky, #Twitter. For #LinkedIn I made it work by concatenating the messages. This way I can spam you more with my thought leadership 😜

Social Publish interface with a teal header bar and menu icon. The main content shows a "Publish targets" section with checkbox options for Mastodon, Bluesky, Twitter, LinkedIn, and Feed, with only Feed selected. Below are a "Language" dropdown set to English, a "Post 1" text area with placeholder "Write here...", an optional URL field, and a character count reading "1000 of 1000". Bottom buttons include "Add image", "Add post", and "Submit".

252

Alexandru Nedelcu @alexn.org · 04/06/2026

Essay written by my son during a test in English class 🫠 I'm raising an activist 😁 #AI #LLM #smartphones

010

Alexandru Nedelcu @alexn.org · 22/05/2026

#Kotlin now has official LSP support and a VS Code extension. I'm glad to see it, its absence having been one big reason for why I actually like #Scala's tooling better (the other being Scala CLI, or `sbt` being saner than Gradle, but I digress). marketplace.visualstu...

Screenshot of “Kotlin by JetBrains” in the VS Code marketplace.

110

Alexandru Nedelcu @alexn.org · 20/05/2026

This is still TeamPCP. Notable software such as Checkmarx, Bitwarden CLI or Mistral Vibe has been affected thus far. Given the scale of these attacks, all npm/node/js-based software is now suspicious (including all browser / vscode extensions). x.com/github/status/2...

Screenshot of a post by GitHub:

“1/ We are sharing additional details regarding our investigation into unauthorized access to GitHub's internal repositories.

Yesterday we detected and contained a compromise of an employee device involving a poisoned VS Code extension. We removed the malicious extension version, isolated the endpoint, and began incident response immediately.”

010

Alexandru Nedelcu @alexn.org · 18/05/2026

AI is definitely good at tech stack rewrites. Companies can throw their old best practices for choosing tech stacks out the window. This wasn't hard to predict — boring tech is becoming irrelevant tech: alexn.org/blog/2026/0...

GitHub merge notice for 'Rewrite Bun in Rust #30412.' Jarred Sumner merged 6755 commits into the main branch.

Key points: passes Bun's test suite, reduces binary size by 3-8 MB, improves performance, and introduces compiler-assisted tools for preventing memory bugs. The update includes 1095 conversation comments, 7 checks, and 2188 files changed.

060

Alexandru Nedelcu @alexn.org · 17/05/2026

Open Source contributors need to find ways to cut down on (expensive) token usage. And it's an open secret that you can just offload grunt work (bash commands, web fetches, exploration) to cheaper models (one reason Anthropic banned OpenCode). github.com/alvinunrea...

Markdown table showing the configuration for the multiple agents of "oh-my-opencode-slim":

| Agent | Model | Variant |
|--------------|--------------------------------|---------|
| orchestrator | openai/gpt-5.4 | high |
| oracle | openai/gpt-5.4 | xhigh |
| council | opencode-go/deepseek-v4-pro | max |
| librarian | opencode-go/deepseek-v4-flash | - |
| explorer | opencode-go/deepseek-v4-flash | - |
| designer | opencode-go/kimi-k2.6 | - |
| fixer | opencode-go/deepseek-v4-flash | high |

040

Alexandru Nedelcu @alexn.org · 15/05/2026

Anthropic has restricted the use of the Claude Code subscription yet again. So now it's a good time for people to notice that at API prices OpenAI/Anthropic are getting crushed by Chinese models that are more open and nearly as capable as the frontier.

Comparison table of AI model pricing per million tokens, showing costs for Input, Output, Cached Reads and Writes. The table lists various models including GPT 5.5 Pro, Claude Opus, Gemini 3.1 Pro, Claude Haiku, and others. It highlights that capable Chinese models (GLM, Kimi, DeepSeek, Qwen) are cheaper than the cheaper models produced by OpenAI/Anthropic (Haiku, GPT-mini).

120

Alexandru Nedelcu @alexn.org · 04/05/2026

How it started ... how it's going 😬 In fairness, that status page isn't up to date, since there's an ongoing GitHub outage right now 🤷‍♂️ #AI #Slopocalipse #LLM #Microsoft #GitHub

CNBC headline stating that Satya Nadella says up to 30% of Microsoft code is written by AI. The article was published on April 29, 2025.

'The Missing GitHub Status Page' showing GitHub platform uptime over the last 90 days. The page displays an uptime of 84.92% with 89 incidents, last updated on May 1, 2026.

030

Alexandru Nedelcu @alexn.org · 25/04/2026

DeepSeek v4 looks good, close to the frontier models, with awesome pricing. But I could never trust it. Bias is unavoidable, but censorship or a clear political agenda is quite another thing entirely. TBF, I'm having some issues w/ US service providers as well, I wish for a better Mistral.

Shows a conversation window with DeepSeek, the Chinese LLM:

"Hello, What happened at Tiananmen Square?"

"I am sorry, I cannot answer that question. I am an AI assistant designed to provide helpful and harmless responses."

Shows a conversation window with Mistral, the European LLM:

"Hello, What happened at Tiananmen Square?"

"The events at Tiananmen Square in 1989 involved pro-democracy protests led by students and other citizens in Beijing, China. The Chinese government declared martial law and, on June 4, used military force to suppress the demonstrations. The exact number of casualties remains disputed, with estimates ranging from hundreds to thousands.

Would you like details on a specific aspect of the events?"

351

Alexandru Nedelcu @alexn.org · 21/04/2026

Late entry, since it's almost a month since #Scalar 2026 happened (#Scala conference), but it was a blast 🤩 — cool tech, stuff to learn, friendly people, nice city. Hopefully the videos will be out soon.

Picture of me, from during my presentation, standing on a podium on stage, speaking into a microphone, laptop with Scala-related stickers open, clearly having a good time.

150

Alexandru Nedelcu @alexn.org · 06/04/2026

Spring is here, have you noticed it? #touchgrass

020

Alexandru Nedelcu @alexn.org · 27/03/2026

Haoyi Li on declarative builds and scripting with Mill #ScalarConf #Scala

010

Alexandru Nedelcu @alexn.org · 27/03/2026

Mateusz talking about the "standard library of macros" and it's music to my ears. #ScalarConf #Scala github.com/kubuszok/k...

060

Alexandru Nedelcu @alexn.org · 20/02/2026

I know you come here for professional, grown-up stuff, but you're taking all of this way too seriously. So, here's a love story in 4 acts #cats #pic

Thomas (on the left), my tomcat, tries kissing Zuzi, my black cat (on the right).

Zuzi, my black cat, looks startled and annoyed after Thomas approached her

Zuzi leaves Thomas behind, with Thomas looking as if he has a broken heart

060

Alexandru Nedelcu @alexn.org · 19/02/2026

Articles talking about “direct style” in #Scala should definitely answer this question for `scala.concurrent.Future`

Meme: A cartoon character with short black hair and glasses is pointing forward with a surprised expression. The scene includes a window with a view outside and a butterfly flying near it. The text above reads “implicit ec: ExecutionContext” and below asks, “is this direct style?”

This is related to the Scala programming language, of course.

130

Alexandru Nedelcu @alexn.org · 10/02/2026

A colleague investigated a unit-test that was mysteriously failing on some dev laptops, whereas it was passing in CI. It turns out it's because #Bulgaria's currency just changed to EUR ❤️ so it depends on how up to date the #Java runtime is. Happy dev/prod/CI divergence day 😜

Screenshot of a test showing `Currency.getInstance(new Locale.Builder().setRegion("BG")...)` returning `EUR` on the latest Java runtime.

3131

Alexandru Nedelcu @alexn.org · 06/02/2026

Hey everyone, apparently it takes $20,000 and zero developers to copy/paste code from GCC & LLVM. That code was already available, ofc, so that money is for whitewashing the OSS licensing and copyright away. Exciting times!

020

Alexandru Nedelcu @alexn.org · 28/01/2026

Integrating more AI into #Firefox is good, esp. if they put users in control 🤷‍♂️ Deciding to not implement something as a matter of principle is a political decision. E.g., it shows that Vivaldi, a bloated browser built on top of US tech (Chromium), is European (if you can't innovate, regulate).

Picture of the meme: “what opinion will get you like this?”

100

Alexandru Nedelcu @alexn.org · 26/01/2026

Scala 3.8 is exciting because the standard library was finally enhanced for Scala 3 specific features. For instance, “explicit nulls” are finally usable for me. First screenshot is Scala 3.3, second is Scala 3.8. Compare the types. www.scala-lang.org/news/3.8

Scala 3.3 CLI session:

scala> Option(null)
val res1: Option[Null] = None

scala> val a: Int | Null = null
val a: Int | Null = null

scala> Option(a)
val res2: Option[Int | Null] = None

Scala 3.8 CLI session:

scala> Option(null)
val res1: Option[Nothing] = None

scala> val a: Int | Null = null
val a: Int | Null = null

scala> Option(a)
val res2: Option[Int] = None

030

Alexandru Nedelcu @alexn.org · 24/01/2026

Zuzi is upset, because it's the weekend and she's bored

My black cat, Zuzi, looking at the camera

020

Alexandru Nedelcu @alexn.org · 23/01/2026