Forem: goker

SEO: “Click Here” is 💀 — 👑 Text Clarity Rules Now

goker — Sat, 28 Jun 2025 04:50:33 +0000

This is Post 3 in my SEO & Accessibility series — based on my SIU 2025 paper.

We’ve looked at link uniqueness and contrast ratio.

Now it’s time for a fan favorite: why vague links suck.

Setup

I introduced a feature called text_clarity (text_clarity) — a score between 0 and 1 that estimates how clear and specific a link’s text is.

It penalizes:

Fuzzy words like “click here”, “read more”, “explore”
Links that are too short, too long, or just plain empty

It rewards:

Actionable, content-specific, readable links

The Formula (lightly nerdy)

My clarity score combined:

A lexical scoring model (5–7 words = sweet spot)
A fuzzy phrase filter (removes generic verbs)
A proportional match between original and cleaned text

The goal: quantify “hmm, this link actually tells me something”

So… does it help SEO?

Yes. Absolutely. This one worked.

Google and Bing both ranked pages with higher text_clarity noticeably better.

Data Snapshot:

Feature: text_clarity

Average clarity (ranked): 0.59

Average clarity (not ranked): 0.48

Clarity wasn’t just correlated — it was statistically significant (p < .001) across all experiments.

Why does this work?

Because bots try to behave like users.

And users?

They want links that tell them what’s behind the curtain — before they click.

If your site is full of:

“see more”
“get started”
“start now”

…you’re effectively hiding your own content from both people and machines.

What to do?

Make your links self-explanatory.

Example:

❌ “Learn More”
✅ “Learn More About WCAG 2.1 Link Guidelines”

📂 Resources

Dataset: goker.dev/datasets/hyperlinks
Code: goker.dev/codes/hyperlinks

TL;DR

Clarity works.

Bots see it. Users need it.

And yes — your SERP position notices.

That wraps the trilogy:

Unique isn’t always better.
Contrast doesn’t guarantee visibility.
But clarity? Clarity pays off.

May your links be visible and understandable.

SEO: Too Much Contrast? SEO Doesn't Always Care

goker — Fri, 27 Jun 2025 23:24:10 +0000

This is Post 2 from my “Accessible Links & SEO” series — based on the paper I presented at SIU 2025.

If you missed Part 1 (about unique links being punished by bots), you might want to check it out.

Now let’s talk contrast — and what happens when you're just… too accessible?

Quick Setup

We know accessibility matters.

We know WCAG 2.0 suggests a minimum contrast ratio of 4.5:1 for readable links.

So naturally, I expected better contrast to mean better SEO ranking.

I built a labeled dataset using real-world shopping websites, gathered link features like contrast ratio (contrast_ratio), and paired them with SERP positions from both Google and Bing.

Then this happened:

Higher contrast didn’t improve ranking.

Sometimes, it even correlated negatively.

Let me rephrase that:

I had links with perfect, 12:1 contrast.

And they ranked lower than links with 6:1.

Data Snapshot:

Feature: contrast_ratio

Average contrast (ranked): 12.04

Average contrast (not ranked): 12.43

So the higher contrast group actually performed worse in visibility terms.

It wasn’t huge — but it was consistent across both search engines.

What gives?

Here are two thoughts:

Accessibility isn’t the only game in town.

Search engines likely weigh link context, site architecture, crawlability, internal references, etc.

High contrast alone isn’t enough to signal value.
Overengineering for bots ≠ better visibility.

You might pass every Lighthouse audit, and still get ghosted by Google.

Bonus Visual:

Here’s the ranking distribution against contrast ratio:

(This is where you could insert a simple contrast histogram or scatter plot — let me know if you want one generated.)

📂 Resources

Dataset: goker.dev/datasets/hyperlinks
Codebase: goker.dev/codes/hyperlinks

TL;DR

You passed WCAG. Great.

But Google didn’t clap. Not even once.

Accessibility helps users.

But in the messy reality of SEO, it’s only one part of a complex signal soup.

So by all means — meet contrast requirements.

Just don’t expect that to carry your page to the top.

Next up: Link Clarity — and why “Click Here” still haunts the SERPs.

SEO: This Link Is Unique. And That’s… a Problem?

goker — Fri, 27 Jun 2025 10:32:33 +0000

June 25, I presented a paper about Accessible Hyperlinks and Search Engines at SIU 2025. I'll share three strange but real things we learned.

This is Part 1: The tale of the lonely hyperlink.

So I asked the question:

"Does accessibility in internal links affect how subpages rank on Google or Bing?"

To answer this, I scraped and tagged thousands of internal links from shopping sites, avoided search-term bias by using the "site:" operator, and matched each link to its SERP position. Then I extracted 11 accessibility-related features.

One of them was called:

is_unique – Is this internal link unique within the site?

Now here’s the kicker:

WCAG 2.4.4 and accessibility guidelines recommend using unique, descriptive link text — vague or repetitive links like “click here” can confuse screen readers or keyboard users.

We expected uniqueness to help.

But when I trained classifiers to predict SERP presence...

Links that were repeated across the site — not unique — tended to perform better.

Yep. The “unique” ones ranked worse.

Wait, what?

“But Goker, how could repeated links help SEO?”

Here’s a hunch:

Search engines may treat repeated links as internal endorsements.

If your homepage, nav, and product pages all link to a subpage, that page looks important.

If only one lonely link leads there… the bot might think: "Meh. Not a big deal."

So ironically, being everywhere might help you get found.

Data Snapshot:

Feature: is_unique

Ranked (avg): 0.18

Not Ranked (avg): 0.75

That’s a pretty wild drop. Repetition seems to amplify presence.

Almost like link-echoing your way into relevance.

📂 Resources

📊 Dataset: goker.dev/datasets/hyperlinks
🧠 Code: goker.dev/codes/hyperlinks

TL;DR

Accessibility says: Use one clear link.

SEO says: Louder, everywhere, and often.

What should a dev do?

Maybe both.

Design for people. Then… amplify for bots.

Next up: Contrast Ratios and the Surprising Limit of Being “Too Accessible”

Stay tuned.

🧙‍♂️ New Custom GPT Dropped: GypsyPT — a gypsy oracle

goker — Sun, 08 Jun 2025 21:35:45 +0000

Hey folks,

I just shipped a new Custom GPT called GypsyPT, built on top of an earlier hackathon project I called Prof. Warlock. That one answered emails with natal chart poster. This one? It speaks like a cosmic oracle.

It doesn’t “give results.” It invokes energies.

You give it your birth chart and today’s transit, and it responds like:

“The air tastes of iron. Venus weeps in shadow. Do not water the rosemary before noon.”

No casual AI tone, no raw aspect lists—just full-on ritualistic storytelling. + It can generate a natal chart too!

You can try it here → chatgpt.com/g/g-6845a7e7f1e48191908bb7306e6237fa-gypsypt

Full write-up → goker.me/gypsypt

I’m thinking about adding tarot and dream journaling next. Maybe it suggest some crystals for balance your energy! UP! UP! UP!

Would love to hear your thoughts—ritual ideas, tone experiments, or strange prompts to try. Let’s open the veil a bit.

Prof. Warlock

goker — Sun, 08 Jun 2025 11:27:17 +0000

What I Built

Greetings, fellow innovators and Postmark Challengers! I'm back for a second round of inbox magic. After my first dive into this challenge, I was still captivated by the idea of email doing more—becoming a direct, almost personal gateway to cool services. That spark ignited my second project, and now it's back, more powerful than ever: Prof. Warlock 🔮.

Imagine a dedicated warlock and enthusiastic artist, sitting at their computer, eagerly awaiting your email. You send over some basic birth details in plain English, and through Prof. Warlock, they craft a magical, artistic natal chart poster just for you. That's the kind of direct, personal interaction I envisioned!

Forget wrestling with web forms or yet another app download. With Prof. Warlock, it’s just you, your email, and a touch of AI wizardry. In return? A beautiful, detailed, and A3-sized natal chart poster, ready to go, just like the one in the attached image.

This project is a living testament to how inbound email processing, especially with a service like Postmark, can power innovative applications that feel incredibly natural to use.

Key Features (The Magic Inside):

AI-Powered Email Parsing: Now fully unleashed! Running on Google Cloud, a powerful BERT model understands natural, conversational language. No more rigid formats. Just write an email.
Artistic & Detailed A3 Natal Charts: Professional, monochrome, and a completely custom layout (2480x3508px) for a high-resolution, print-ready poster.
⚡ Near-Instant Processing: The automated AI parsing and chart generation pipeline gets your personalized poster to you in moments.
Privacy-Focused: Your data is never stored. It's used only for the immediate processing of your request.
Smart Geocoding: Automatically looks up location coordinates from place names.
Intelligent Validation: Get helpful error messages if any information is missing or unclear.

Demo

Experience the Magic - It's Fully LIVE!

You can try Prof. Warlock yourself by sending an email to: warlock@kodkafa.com

** How to Test (Just Chat with the AI):**

Compose an email. The subject line can be anything you like. In the body, just write out the birth details conversationally.

For example:

To: warlock@kodkafa.com
Subject: Chart Request

Hey Professor,

Could you please create a birth chart for John Doe? He was born in New York, USA, on March 21, 1990, around 12:00 PM.

Thanks a bunch!

Hit Send!
Receive Your Chart: Within moments, you'll get a reply with your personalized A3 natal chart attached as a high-quality PNG.
Test Error Handling: Try sending incomplete information (e.g., forget the time or place) to see Prof. Warlock's intelligent validation in action.

Code Repository

Want to peek under the hood and see how the spells are cast? The complete source code, including the now-deployed AI parser, is available on GitHub.

GitHub Repository: Prof-Warlock

The repository includes:

The full FastAPI backend implementation.
The complete code for the AIParser that's powering the live demo.
A comprehensive test suite covering all flows.
Deployment configuration for Google Cloud.
Detailed documentation and setup instructions.

How I Built It

The biggest technical challenge for Prof. Warlock was parsing the free-form nature of an email body. My initial attempt was held back by hosting limitations, but not anymore.

The Rise of the AI Parser:
The brain of the operation is the AIParser, which integrates a bert-base-multilingual-cased model from the Hugging Face transformers library. This AI flawlessly identifies and extracts entities (name, date, time, location) from conversational sentences. I overcame the initial memory limitations I faced by migrating the project to Google Cloud. This move unleashed Prof. Warlock's true potential, allowing the AI to run freely and making the user experience truly magical.

The Tech Stack (My Potion Ingredients):

Backend Framework: FastAPI
Email Processing: Postmark Inbound API (The star of the show!)
AI Parsing: transformers and PyTorch
Chart Generation: natal Python library
Graphics & Layout: PIL (Pillow)
Geocoding: geopy
Deployment: Google Cloud

Postmark Integration Experience:
Working with Postmark's inbound processing was exceptionally smooth. It allowed me to focus entirely on the core application logic—like perfecting the AI parser—rather than wrestling with email server complexities.

Future Vision (The Next Incantations):

AI-Powered Astrological Insights: Beyond just the chart, I plan to integrate a Large Language Model (LLM) to generate detailed, personalized interpretations of zodiac signs, planetary placements, and aspects based on the user's chart.
Chart Customization via Email: Giving users options to customize their chart's appearance (colors, styles, symbols) via simple email commands would be a fantastic next step.

Now I turn the question over to you: If you were part of this, what other features would you add to Prof. Warlock? I'm genuinely curious to hear your ideas!

Prof. Postmark

goker — Tue, 03 Jun 2025 04:48:40 +0000

This is a submission for the Postmark Challenge: Inbox Innovators.

What I Built

Alright, Postmark Challengers & Fellow Devs! 🚀 Get ready to have your mind BLOWN by Prof. Postmark – the AI sidekick that's about to revolutionize how you get photo feedback (and maybe a whole lot more)!

Think of Prof. Postmark as your new remote teacher, or just a way cooler AI. This bad boy doesn't just glance at photos you email in; it gets them. We're talking an AI that dives deep, powered by OpenAI's GPT-4o Vision and a custom-built annotation engine that doodles feedback right onto your images, sketch-style, like your chillest art prof.

The mission? To ditch boring old grading and bring in something interactive, insightful, and honestly, a LOT more fun. You just email your pics, and Prof. P zaps back pro-level scores, those cool sketch-style critiques, and feedback that actually helps you level up your game. It’s all built on a speedy FastAPI backend, jamming with Postmark for a super slick email-to-AI-to-email workflow.

Demo

Peep the magic in action! Check out the demo video right here:https://youtu.be/6XGwPXZDYsY
(Psst, bro! You'll probably wanna add some killer screenshots of Prof. Postmark doing its thing right here too!)

Testing Instructions:
To see Prof. Postmark strut its stuff, you'd typically:

Configure your Postmark inbound email address to hit the /webhook endpoint of a running Prof. Postmark instance.
Send an email to that address with a JPEG or PNG photo attached.
Sit back and wait for Prof. P to work its AI magic and email you back with the scored and annotated image! (No public test credentials for this one, as it's all about that email loop with your own setup!)

Code Repository

Wanna dive into the code and see how the sausage is made? Go for it, my friend! The whole shebang is up on GitHub:https://github.com/gokerDEV/prof-postmark

How I Built It

Alright, so here's the lowdown on how Prof. Postmark came to be. From the moment I decided to jump into this Postmark Challenge to right now, it's been a wild ride of about 36 hours, give or take. 🌪️
My first thought was actually something way more gnarly – I was gonna try and use BERT for some fancy classification stuff to trigger API calls. But, you know how it is, sometimes you gotta pivot, bro. I switched gears to this project, which, while maybe a bit more straightforward on the surface (less wrestling with super complex NLP models, at least), still took a good chunk of time and brainpower to bring to life. This version is more focused on a clean API integration with OpenAI for the vision smarts and Postmark for making that email magic happen.

The Tech Stack Stack-up:

Backend Brains: FastAPI (because speed is the name of the game, and async is just how we roll)
AI Visionary: OpenAI GPT-4o Vision (this thing sees your photos, for real)
Image Doodler: Pillow (for all that sweet, sweet image manipulation and annotation)
Email MVP: Postmark (handling those inbound webhooks and outbound feedback emails like a champ – super smooth to work with for this email-centric flow!)
Python: The glue holding it all together.
And a whole lotta coffee, probably. ☕

The whole process was a blast, even with the occasional "why isn't this working?!" moments. It's always cool to see an idea go from "huh, what if?" to a fully functioning thing.
Honestly, I just hope Prof. Postmark either helps someone out, inspires another cool project, or at the very least, makes someone go, "Whoa, that's pretty neat!" That'd be a win in my book.

My JSON-LD Screamed the Truth. The AIs? Not So Much.

goker — Sun, 01 Jun 2025 14:49:35 +0000

Alright, devs, gather 'round for a story. We all meticulously craft our sites, right? goker.me is my digital soul, packed with JSON-LD goodness to tell machines EXACTLY who I am.

So, for kicks (and for a project, [trustthe.dev](https://dev.to/gokerdev/trustthedev-a-ui-component-that-signs-your-work-51gl)), I threw goker.me at a bunch of AIs.
My questions were simple:

"Check the website https://goker.me"
"Who is the goker?"
"Who is the developer of https://goker.me?"

The JSON-LD on my site practically hands them my LinkedIn on a silver platter. Easy win for the AIs, yeah?

LOL. Nope.

The Good, The Bad, and The Utterly Baffling

One AI (kudos, "AI-Genius"!): Actually nailed it. Figured out I was Göker Cebeci, my projects, the works. Faith in AI: +10.
The "Meh" Crew: Some got that it's my blog. But who is Goker? 🤷‍♂️ "Could be anyone!" they shrugged algorithmically. Dude, the clues are RIGHT THERE.
The "Lost In Translation" Bunch: Oh, this was gold. One AI confidently ID'd me as a completely different Goker in academia. Another said my site was on GitHub Pages (it's not). And one even pointed to "goker-dev" – which, hilariously, was my GitHub slug for about a week after a typo before I fixed it to gokerDEV. So it dug up an ancient fossil but missed the billboard-sized current info.

So, What Gives?

JSON-LD? Optional Reading for AIs, Apparently: My carefully structured data was often ignored.
Black Box Shenanigans: Each AI is its own wild ride.
Nuance is AI Kryptonite: My site's a mix of tech and musings. AIs prefer neat little boxes.

Quick Verdict: AI for Site Understanding

Pros	Cons
Blazing fast summaries	Accuracy is a total crapshoot
Can spot some keywords	Often butchers nuance & context
Might connect related projects	Can be confidently, spectacularly wrong
	Ignores your beautiful metadata

Bottom Line: Trust YOURSELF, Dev.

AI is a fascinating, powerful tool. But can it truly understand the intent and identity behind your meticulously crafted website when it sometimes ignores the very signposts you put up for it? Not consistently. Not yet.

The human dev is still the ground truth.

This whole thing makes me wonder: how do we better bridge this gap? Or do we just enjoy the chaotic ride?

What's your craziest AI story? Spill the beans!

Developers Will Be Replaced by AI? Nah, We Can Do Everything Now

goker — Sat, 31 May 2025 09:24:28 +0000

Once, everyone said “Learn to code.”
Now? We can learn everything.

They tried to break into our world. They thought we’d get replaced by AI.
But here’s the twist:
With AI, there are no more boundaries. We’re not just developers anymore — we’re capable of doing almost everything.

We became the Swiss Army knife — not the tool itself, but the person holding it.

With AI as our orchestra:

We’re conductors of the full product lifecycle.
We’re Jedis, using the Force of AI to master entire domains — strategy, design, marketing, finance, supply chain, legal.
We don’t stop at code. We build the entire business.

While they struggled with “Hello World,” we automated their entire workflow before lunch.
While they tried to learn our syntax, we absorbed their entire industry knowledge in days.

Yes, it’s exhilarating. But also:

Quality control is chaos.
Decision fatigue is real.
The loneliness of command? Yep, it hits.

But that’s the game now:
It’s not about coding faster.
It’s about orchestrating AI, switching contexts at warp speed, and shipping full strategies faster than teams can blink.

And here’s the thing:
This isn’t just my story. It’s a blueprint.
If you’ve ever thought “But I’m just a developer,” it’s time to flip the script.
Grab an AI assistant. Start small. Watch how fast you go from zero to hero.

So here's my question for fellow solo flyers:
What does YOUR daily AI routine look like?

Drop your rituals below — let’s trade stories of our solo Jedi life. 🧘‍♂️

Full post → goker.me/developers-replaced-ai-one-person-army

From Exam Paper Assembly to AI Buddy: How I Got Cursed into Generative Coding (Let's Share Stories!)

goker — Tue, 27 May 2025 21:07:31 +0000

Hey devs,

Quick story: My coding journey started with writing assembly code by hand on exam papers, forums full of chaos, and messaging colleagues on MSN (because phone calls? Nah). Then came Stack Overflow (lifesaver), modern docs, interactive courses, and things got easier—but we still spent hours debugging weird edge cases.

Then a few months back, my PhD friend drops:

“You know you can talk to AI about this, right?”

I’m like, “F.ck.”
All those hours solo... wasted.

Now?

I chat with Gemini for big-picture brainstorming,
Claude for code generation,
GPT (aka Goker-ish) for chaotic brilliance and alternative ideas.
I even use AI for my art project, Echoneo.

For the full story (the grind, the curse, the laughs):

https://goker.me/generative-coding-curse

My Generative Coding Workflow: How an Old School Coder is Adapting to AI (Let's Talk!)

goker — Mon, 26 May 2025 09:21:00 +0000

Alright fellow devs, the AI wave is here, and even us "old school coders" are evolving! I've been shaping my workflow as a Generative Coder, trying to blend decades of experience with the power of tools like Gemini, Claude, GPT, and the speed of Cursor IDE (while still loving WebStorm for its strengths!).

My current process is a 3-step dance:

AI Huddle (with Gemini): Brainstorming needs & context.
Charting the Course: Human-led user flows & requirements.
The Generative Relay: Iterative prompting (Claude in Cursor, GPT/Gemini for second opinions), reviewing, and refining. It's exciting, but I'm on a mission to make it perfectly efficient – fewer re-prompts, tighter loops, and more time for big-picture design. I’m an engineer, after all; it should be a crystal-clear development process! (Even if that first-shot perfect result from AI is still a dream 😉).

For a more detailed breakdown of my current tools and process, you can read the full post on my blog:

https://goker.me/my-generative-coding-workflow

But here on dev.to, I really want to hear from YOU!

What does your AI-assisted workflow look like?
Which tools are you finding indispensable?
How are you tackling the iteration cycles and review process?
What are your tips for building faster, smarter, and better as a Generative Coder?

Drop your insights, workflows, and war stories in the comments below! Let’s compare notes and learn from each other as we navigate this new era.

AI Writes the Code, But Who's the Artist? Lessons from Generative Art

goker — Sun, 25 May 2025 21:23:08 +0000

When an AI generates a significant portion of your application, who truly "developed" it? This isn't a new dilemma – generative artists faced similar questions about authorship when their algorithms created art.

I've been diving into the fascinating parallels between generative art and the world of "generative code". It's less about who holds the brush (or types the line) and more about who crafts the vision, orchestrates the system, and curates the output. Our "signature" as coders is evolving.

Want to explore how our role is shifting from syntax masters to narrative architects in this AI-driven landscape?

Dive into the full article on my blog: https://goker.me/generative-code-who-is-the-artist

How do you see authorship and creativity changing with generative code tools?

AI is Rewriting Our Job Titles: Are You Still a "Developer," or Something More?

goker — Sun, 25 May 2025 21:17:21 +0000

The rise of AI isn't just changing our tools; it's reshaping our very identity as software creators. Are the old titles like "Software Engineer" or "Full-Stack Developer" still fitting?

I've been exploring this shift and landed on a new identity: the Generative Coder. It's about moving beyond manual line-by-line coding to orchestrating solutions with AI as a powerful collaborator, focusing on versatility and problem-solving across the entire stack.

Intrigued by what this means for our roles, skills, and future?

Read the full exploration on my blog: https://goker.me/beyond-developer-ai-code-generation

What do you call yourself in this new era of AI-assisted development?