<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jubiar Ahmed on Medium]]></title>
        <description><![CDATA[Stories by Jubiar Ahmed on Medium]]></description>
        <link>https://medium.com/@mdjubiarahmedrabbi?source=rss-f093fd46f604------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*xW4F_CzEXZvo0Rmcjr5XTg.jpeg</url>
            <title>Stories by Jubiar Ahmed on Medium</title>
            <link>https://medium.com/@mdjubiarahmedrabbi?source=rss-f093fd46f604------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 20:47:04 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mdjubiarahmedrabbi/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[You Got What You Wanted. So Why Does It Feel Like Nothing?]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/you-got-what-you-wanted-so-why-does-it-feel-like-nothing-e93559cdd5a0?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/e93559cdd5a0</guid>
            <category><![CDATA[personal-growth]]></category>
            <category><![CDATA[social-media]]></category>
            <category><![CDATA[mental-health]]></category>
            <category><![CDATA[happiness]]></category>
            <category><![CDATA[life-lessons]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Mon, 17 Aug 2026 04:30:47 GMT</pubDate>
            <atom:updated>2026-08-17T04:30:47.512Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2917ojdQTCauOszMZjDNwg.png" /><figcaption>We spend years climbing toward the thing we’re sure will feel like <em>arriving</em> — and get there only to think, <em>huh, is this it?</em></figcaption></figure><p>Nobody warns you about this one.</p><p>We spend years chasing a thing — the trip, the followers, the milestone, the promotion — completely sure that arriving will feel like <em>arriving</em>. And then we get there, stand in the middle of the exact thing we wanted, and quietly think: <em>huh. Is this it?</em></p><p>It’s the twist at the heart of modern life. We are brilliant at getting what we <em>think</em> we want, and terrible at noticing it was never quite what we needed. The two things look almost identical from a distance — same shape, same shine — and it’s only up close, holding the thing, that you feel the difference. One fills you up. The other just looks like it should.</p><p>Let me show you three places this quietly happens: the trip, the phone, and the stranger who’s about to make all of your worst decisions.</p><h3>The harder you chase happiness, the faster it runs.</h3><p>Picture the “perfect” vacation. Planned to the minute. Every sunset scheduled to be breathtaking, every dinner obligated to be the best meal of your life.</p><p>And you spend half the trip anxious that it isn’t magical <em>enough</em>.</p><p>That’s the happiness paradox in a single image. We treat joy like a destination we can reach if we just optimize the route well enough. But the moment you start checking — <em>am I happy yet? should this feel better? am I wasting this?</em> — you’ve already stepped outside the moment that would have made you happy in the first place. Happiness is a bit like sleep: the harder you try to force it, the more stubbornly it dodges you.</p><p>Why do we keep falling for it? Because somewhere along the way, someone convinced us that happiness is a score to maximize — more experiences, more highlights, more proof. But feelings aren’t a scoreboard. They’re weather. You don’t <em>achieve</em> a good mood; you create the right conditions and let it roll in on its own.</p><p>So stop auditing your own joy. Lose yourself in a conversation, a walk, a task you actually enjoy — and let the good feeling ambush you from the side. It almost always does, right at the moment you finally stop checking for it.</p><h3>You are more connected than any human in history — and possibly lonelier for it.</h3><p>Here’s a small, brutal test. It’s 2 a.m. Something’s genuinely wrong. Who do you call?</p><p>Now scroll your followers. Your group chats. The hundreds of names that “like” your life. And notice how few of them survive that one question.</p><p>That’s the paradox nobody posts about. We have infinite ways to reach people and less closeness than ever, because likes are the empty calories of connection — they taste like something and feed you nothing. A heart on your photo is not a hand on your shoulder. A hundred “happy birthday!” comments is not one person who remembered to actually call.</p><p>The reason is quietly sad: online, we broadcast our highlights. So we end up intimately familiar with everyone’s best moments and total strangers to anyone’s worst ones. We know what our old classmate had for brunch and have no idea that they’re barely holding it together. Endless contact. Almost no real contact. It’s possible to spend all day “with people” and go to bed feeling like nobody actually saw you.</p><p>The move isn’t a bigger audience. It’s a smaller, realer circle. Text one person tonight — not “hey,” but “how are you <em>actually</em> doing?” Call the friend instead of liking their post. One honest, slightly awkward conversation outweighs a thousand tidy thumbs-up.</p><p>So — who’s your 2 a.m. person? And when did you last make sure they know they’re yours?</p><h3>Your future self is a total stranger, and you keep throwing them under the bus.</h3><p><em>“I’ll start exercising next month.” “I’ll save later.” “Future me can deal with it.”</em></p><p>Meet the most abused person in your life: you, five years from now.</p><p>Here’s the genuinely strange part — this isn’t just a convenient excuse, it’s roughly how the brain is wired. When you picture your future self, your brain tends to light up the way it does for a <em>stranger</em>, not for you. Which means every time you say “future me will handle it,” some quiet part of your mind hears “not my problem.” So of course you keep spending their money, skipping their workouts, and booking the dentist appointment they’ll have to survive. It’s genuinely hard to sacrifice today for someone who feels like a different person entirely.</p><p>That’s why saving, studying, or building a habit feels so joyless right now. You do all the sweating; some stranger with your face collects the reward. The effort is yours, the payoff is theirs, and the two never seem to meet.</p><p>The trick is to make that stranger real. Don’t picture “the future” as a vague fog — picture <em>them</em>. The specific you at 35, or 50, opening the results of every choice you’re making today. Would they thank you, or quietly sigh? Then do one small, kind thing for that person right now. A single walk. Ten dollars saved. One honest text you keep meaning to send. They are not as far away as they feel — and they remember everything you do today.</p><h3>The pattern hiding under all three.</h3><p>The vacation. The follower count. The “I’ll start Monday.”</p><p>Every one of them is us reaching for the shiny version of a real need — and coming home with the wrong thing. We wanted joy and grabbed pressure. We wanted closeness and collected contacts. We wanted a good life and kept mailing the bill to a stranger who happens to share our name.</p><p>The fix isn’t wanting <em>less</em>. It’s wanting more <em>honestly</em>. Feel your happiness instead of performing it. Trade a little visibility for a little intimacy. Treat tomorrow’s you like someone you love — because, whether it feels like it or not, you do.</p><p>Because the life you keep chasing usually isn’t waiting somewhere ahead of you, polished and perfect and finally enough. It’s sitting right here, in the ordinary, un-optimized, slightly imperfect moment you keep sprinting past on your way to something better.</p><p>You’ve probably already had a hundred moments that were exactly enough. You were just too busy reaching for the next one to notice.</p><p>So — what would you do differently today if your future self were already in the room, watching, and quietly rooting for you?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e93559cdd5a0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[You’re Not Lazy — You’re Just Optimized to Death]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/youre-not-lazy-you-re-just-optimized-to-death-8cc881b0613e?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/8cc881b0613e</guid>
            <category><![CDATA[life-lessons]]></category>
            <category><![CDATA[procrastination]]></category>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[self-improvement]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Mon, 10 Aug 2026 05:15:33 GMT</pubDate>
            <atom:updated>2026-08-10T05:15:33.856Z</atom:updated>
            <content:encoded><![CDATA[<p>There’s a specific kind of tired that has nothing to do with hard work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iLYPmC2ZFlrAmhO6bcBF8Q.png" /></figure><p>You feel it around 6 p.m. You’ve been “busy” since morning — replies fired off within seconds, four meetings survived, inbox scrubbed down to a clean, glowing zero. And still, when you shut the laptop, something feels off. Because the one thing you actually sat down to do this morning? It’s still sitting there, untouched, watching you.</p><p>That’s not laziness. That’s the trap.</p><p>We are the most optimized generation in history. We track our sleep, our steps, our screen time. We have systems for our systems, apps that manage our other apps, and a quiet, constant belief that if we could just get a little more efficient, life would finally click into place. And somehow it keeps feeling <em>harder</em>, not easier.</p><p>So here’s the uncomfortable idea I want to hand you: a lot of the effort we pour into improving our lives is quietly making them worse. Not the big, dramatic stuff. The small, invisible stuff we do every single day, convinced we’re helping ourselves.</p><p>Let me show you four ways — and I’d bet money you’ll recognize every single one.</p><h3>Busy is not the same as productive. Your brain just wishes it were.</h3><p>Answering forty messages feels like an accomplishment. It isn’t. It’s forty tiny dopamine hits dressed up as progress.</p><p>The problem is that your brain is a sucker for anything that feels <em>finished</em>. Replying to an email gives you a clean little “done.” The real work — the report, the project, the thing that actually moves your life forward — is slow, foggy, and rewards you with nothing for hours. So your brain, ever the coward, picks the easy wins and lets you call it a full day.</p><p>You didn’t have a productive day. You had a <em>reactive</em> one. You spent eight hours being a very responsive assistant to other people’s priorities, and none on your own.</p><p>The fix isn’t working more. It’s guarding one or two hours for the hard thing <em>before</em> the noise floods in — phone in another room, notifications off, permission to be briefly unreachable. The world will survive you being slow to reply. Your goals might not survive you never getting to them. Do the right little, not the busy lot.</p><h3>The more choices you have, the less you’ll enjoy any of them.</h3><p>You open Netflix to unwind. Twenty minutes later you’ve watched nine trailers, added four things to a list you’ll never open again, and you go to bed having watched exactly nothing.</p><p>Congratulations — you’ve met the choice paradox. We think more options mean more freedom. But every extra option quietly raises the bar for the “right” choice, until picking anything feels like a small failure.</p><p>Here’s the sneaky part: it’s not the choosing that drains you. It’s the <em>mourning</em>. Every option you pick means saying goodbye to all the ones you didn’t — and your brain grieves each one. More doors open means more doors to regret. This is why a tiny café with six things on the menu feels relaxing, and a giant one with two hundred feels like a test.</p><p>So shrink the menu on purpose. Pick from the first five. Decide, before you even start scrolling, that “good and now” beats “perfect and never.” A film you actually watched will always outrank the flawless one you spent the evening searching for.</p><h3>We save our deepest thinking for the least important decisions.</h3><p>You’ll agonize over headphones longer than you agonized over your degree.</p><p>Twenty tabs. A comparison spreadsheet. A YouTube video of a stranger touching them in 4K. All for a $60 purchase that, honestly, would have been fine either way.</p><p>Why do we do this? Because researching <em>feels</em> like doing something, without the risk of committing to anything. Reading the fortieth review is safe. Clicking “buy” means you might be slightly wrong — and being slightly wrong feels scarier to us than it has any right to. Overthinking is just fear wearing the costume of diligence.</p><p>And notice the cruel joke: the decisions that actually deserve deep thought — where to live, who to spend time with, what to do with your one wild year — we often make on a gut feeling in an afternoon. But the headphones? Three days of research.</p><p>Try this instead: for anything low-stakes, set a ten-minute timer, then decide and move on. Save your real thinking for the things that will still matter in five years. You’ll almost never regret the headphones. You’ll just get your evenings back.</p><h3>Your brain loves procrastination. Even the part of you that hates it.</h3><p>You have something important due. So obviously, right now, the most urgent thing in the universe is… reorganizing your sock drawer.</p><p>Here’s what nobody told you: procrastination isn’t a time-management problem. It’s an <em>emotion</em>-management problem. The task makes you feel anxious, or bored, or lost — and cleaning, scrolling, and snacking all offer instant relief from that feeling. You’re not avoiding the work. You’re avoiding how the work <em>makes you feel</em>.</p><p>That’s also why “just have more discipline” never works. You can’t discipline your way out of a feeling. You can only make the feeling smaller.</p><p>So shrink the task until it’s too small to be scary. Don’t “write the essay.” Just “open the file and type one ugly sentence.” Don’t “go to the gym.” Just “put on the shoes.” Motivation is a liar — it shows up <em>after</em> you start, not before. Begin badly, on purpose, and momentum quietly does the rest. The hardest part was never the work. It was the three minutes before it.</p><h3>Here’s the thread tying it all together.</h3><p>The dead inbox. The endless Netflix scroll. The headphone spreadsheet. The essay you’re avoiding by folding socks. They look like four different problems.</p><p>They’re the same one.</p><p>Every time, we’re trying to do life <em>perfectly</em> — and making it heavier in the process. We optimize the fun out of our evenings, the ease out of our choices, and the doing out of our to-do lists. We’ve become so good at managing life that we’ve half forgotten to actually have one.</p><p>So maybe the real skill of modern life isn’t optimizing everything. Maybe it’s knowing what to leave gloriously <em>un</em>-optimized. Watch the good-enough movie. Buy the decent headphones. Write the messy first draft. Let three emails wait until tomorrow. Choose “done” over “perfect” often enough that you start to enjoy the doing again.</p><p>Because a life squeezed for maximum efficiency tends to have the least room left for the one thing it was all supposed to be for.</p><p>So tonight, when you catch yourself hunting for the perfect option — stop. Smile. Pick one.</p><p>Your evening has been waiting for you this whole time.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8cc881b0613e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Selenium vs Playwright: The Duel Every QA Engineer Eventually Fights]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/selenium-vs-playwright-the-duel-every-qa-engineer-eventually-fights-3cf919e1ca61?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/3cf919e1ca61</guid>
            <category><![CDATA[playwrights]]></category>
            <category><![CDATA[selenium]]></category>
            <category><![CDATA[software-testing]]></category>
            <category><![CDATA[test-automation]]></category>
            <category><![CDATA[qa-engineering]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Mon, 29 Jun 2026 05:49:31 GMT</pubDate>
            <atom:updated>2026-06-29T05:49:31.936Z</atom:updated>
            <content:encoded><![CDATA[<p>A story about two browser-automation tools, one flaky test suite, and the choice that quietly decides your next two years.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oR4U4B6BOuzp4ARd7ZBQoQ.png" /><figcaption>Two tools enter. Your test suite decides who wins</figcaption></figure><h3>The 2 AM Phone Call</h3><p>It always starts the same way.</p><p>Your test suite runs green on your laptop. You push to CI. You go home. At 2 AM your phone buzzes — the build is red. Not because the code broke, but because a button <em>“wasn’t clickable yet.”</em> You add a wait. It passes. Next week, a different button. Another wait. Another 2 AM.</p><p>If you’ve lived this, you’ve already met the real villain of automation: <strong>flakiness</strong>. And the two heroes who promise to fight it are <strong>Selenium</strong> and <strong>Playwright</strong>.</p><p>Let me introduce them properly — because choosing wrong here is the difference between sleeping through the night and becoming best friends with your CI dashboard.</p><h3>Meet the Veteran: Selenium</h3><p>Selenium was born in 2004, built by Jason Huggins to kill repetitive manual testing. It became <em>the</em> standard — so dominant that its WebDriver protocol is now a W3C web standard. That’s not a small thing. It means Selenium isn’t just a tool; it’s part of the web’s official furniture.</p><p>What makes it the veteran: it speaks every language your team writes — Java, C#, Python, Ruby, JavaScript. It touches every browser, including the old ghosts like Internet Explorer. Pair it with Appium and it drives real mobile devices. And it carries a 20-year ecosystem — if you have a problem, someone solved it in 2017 and blogged about it.</p><p>The catch? Selenium talks to the browser through WebDriver, which routes commands over HTTP — a translator standing between you and the browser, and translators add delay. Worse, Selenium doesn’t wait for you. You have to tell it when an element is ready. Forget, and you get the 2 AM phone call.</p><h3>Meet the Challenger: Playwright</h3><p>Playwright showed up in 2020, built by Microsoft engineers — several of them ex-Puppeteer people who knew exactly what they wanted to fix. Instead of going through WebDriver, Playwright talks <strong>directly</strong> to the browser over a persistent connection. No translator. No middleman.</p><p>That one architectural decision is the whole story — so it’s worth seeing it:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tTAG2-jWVHvjxIBelY5myQ.png" /><figcaption>Same job, one fewer middleman</figcaption></figure><p>That direct line is why Playwright auto-waits (it checks an element exists, is visible, and is clickable before acting — the number one cause of flaky tests basically disappears), ships Chromium, Firefox, and WebKit in the box with no driver management, and runs roughly 30–50% faster on big suites. Its “browser context” model spins up dozens of isolated sessions in one process, cutting CI cost. It can even mock APIs and block trackers natively, no proxy gymnastics.</p><p>The catch? It’s younger. Its Java and C# bindings exist but get less love than JavaScript and Python. The community, while exploding past 45M monthly npm downloads, is still smaller than Selenium’s two-decade library of answers.</p><h3>The Head-to-Head</h3><p>Here’s the whole fight on one card:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aEPA2h8qmBQdpxYpK9iz3Q.png" /><figcaption>The whole fight on one card.</figcaption></figure><h3>So… Which One? (The Honest Answer)</h3><p>Here’s the part most articles dodge: <strong>there is no universal winner.</strong> There’s only the right fit for <em>your</em> situation. Two questions usually settle it:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Rn-Z0q-aNSwUIKgCIHKt6g.png" /><figcaption>Two questions. One answer</figcaption></figure><p><strong>Reach for Playwright</strong> when you’re starting a new project on modern browsers, testing dynamic SPAs (React, Vue, Angular) where Selenium feels brittle, chasing speed and low flakiness, cutting CI cost, or working in JavaScript, TypeScript, or Python.</p><p><strong>Stick with Selenium</strong> when you’re a Java or C# enterprise shop with mature bindings, you need legacy or uncommon browser support (yes, IE still haunts some banks), you need real native mobile via Appium, or you already own a large, stable Selenium suite — a full migration rarely pays for itself.</p><p>The quiet truth from senior architects in 2026: Selenium isn’t dying — it embraced WebDriver BiDi to close the gap. But for a brand-new modern web app, Playwright has effectively become the default. Both are actively maintained. Both will outlive this article.</p><h3>Best Practices (Whichever Side You Pick)</h3><p>Never hardcode sleeps — lean on Playwright’s auto-waiting, or use explicit waits in Selenium, never Thread.sleep(). Prefer stable selectors like data-testid over fragile CSS or XPath chains; every selector is a future maintenance bill. Keep tests independent so none depends on another&#39;s leftovers — isolation kills flakiness. Follow the Page Object Model to separate <em>what</em> you test from <em>how</em> you find elements. Run in parallel and in CI early, because a test that only passes on your machine isn&#39;t a test, it&#39;s a hope. And don&#39;t migrate just because it&#39;s trendy — switch for a reason, not a vibe.</p><h3>The Moral of the Story</h3><p>Selenium is the seasoned veteran who knows every road — but expects you to do the driving. Playwright is the modern machine that handles the tricky parts itself, as long as the road is paved with modern web.</p><p>The flaky-test villain doesn’t care which hero you pick. It only loses when you choose the one that fits your stack, your team, and your timeline — and then use it well.</p><p>So the next time your phone buzzes at 2 AM, make sure it’s <em>your</em> choice that answers it.</p><p><em>If this helped, drop a 👏 and tell me in the comments: are you Team Selenium, Team Playwright — or quietly running both?</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3cf919e1ca61" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Being Happy Is Not the Reward. It’s the Engine.]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/being-happy-is-not-the-reward-its-the-engine-236d19325f9e?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/236d19325f9e</guid>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[life-lessons]]></category>
            <category><![CDATA[mental-health]]></category>
            <category><![CDATA[self-improvement]]></category>
            <category><![CDATA[happiness]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Tue, 23 Jun 2026 06:30:31 GMT</pubDate>
            <atom:updated>2026-06-23T06:30:31.746Z</atom:updated>
            <content:encoded><![CDATA[<p>Your body has been quietly running on happiness this whole time — you just kept calling it something else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Royv53noXOf1ap5F2EeRaA.jpeg" /><figcaption>Happiness was never the prize at the finish line — it’s the weather you walk through to get there</figcaption></figure><p>Sometimes you’re doing nothing important — making tea, walking somewhere — and for no reason you just feel okay. And then a thought sneaks in: <em>does this even matter?</em> It’s nice, but it passes. So what’s the point of it.</p><p>I used to think of happiness exactly like that. Decoration. The reward sitting at the finish line — do the work, earn the moment, feel good for a bit, back to work. Happiness was the dessert, never the meal.</p><p>I had it backwards.</p><p>It isn’t waiting at the end of anything. It’s upstream. It’s quietly working on you the whole time, and you don’t notice because it doesn’t announce itself the way stress does.</p><p>Researchers noticed that when people feel good, their thinking literally opens up — they see more, connect more, try more. Fear does the opposite: the mind narrows down to the threat. So a good mood isn’t just pleasant. It’s the state where you’re capable of more.</p><p>Then the body stuff got strange.</p><p>Some researchers once scored the autobiographies a group of nuns had written around age twenty-two — just for warmth, how much genuine positive feeling was in the words. Decades later, the warm ones were outliving the rest by years. Same convent, same food, same everything.</p><p>And there’s the one that sounds like a prank. A team measured a few hundred healthy people for everyday positive feeling, then squirted a cold virus up everyone’s nose — same dose for all. The low-positive group got sick about three times as often. The virus went in. Something inside them just answered it differently.</p><p>That’s what gets me. We treat happiness like weather, here and gone. But it reaches into the body — into how long you last, how hard a virus hits. That’s not decoration. That’s load-bearing.</p><p>So the obvious move is: great, go get some. Make it the goal.</p><p>And that’s exactly where it falls apart. The moment happiness becomes the target, it backs away. You finally get the thing you were sure would fix everything, and it does — for about a week — then it’s just your life with a bigger price tag. Chase it head-on and it dissolves.</p><p>So you’re left with a strange shape. The benefits are real — your mind, your body, the people around you. But they only show up when happiness is a state you’re <em>living in</em>, not a prize you’re hunting. It works as a byproduct.</p><p>Which finally made it click. Happiness was never the gold medal for living correctly. It’s good weather for the things you’re actually doing. Nobody lives <em>for</em> a sunny day. But notice how everything grows a little better when the weather’s good.</p><p>So I stopped checking whether I was happy like a score. The better question turned out smaller: <em>is the weather good enough right now to do the thing in front of me?</em></p><p>And when it is, I try not to waste it staring at the sky.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=236d19325f9e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Garbage In, Gospel Out]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/garbage-in-gospel-out-1b53455d5a8f?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b53455d5a8f</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[misinformation]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai-ethics]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Mon, 15 Jun 2026 09:45:14 GMT</pubDate>
            <atom:updated>2026-06-15T09:45:14.406Z</atom:updated>
            <content:encoded><![CDATA[<p>An AI only knows what we feed it — and it never tells you when the food was bad.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MRnNq_nJmV0mH9p3sXScEw.png" /><figcaption>A model is only a mirror of its training data. Feed it a flawed world and it hands that world back — calm, confident, and never flagging that the food was bad.</figcaption></figure><p>We’ve started treating the machine’s answer the way we used to treat a thick reference book on a high shelf: settled, neutral, finished. The doctor’s office runs our case through software to decide how much care we need. The bank runs our application through a model to score our risk. A lawyer asks a chatbot for the right precedent. A student asks it what really happened in 1989. A machine speaks, and we lean in, because it sounds so sure.</p><p>But a model doesn’t <em>know</em> anything. It has never sat with a patient or read an entire legal archive or lived through the history it summarizes. It’s a mirror held up to a vast pile of human data, and like any mirror it can only reflect what’s placed in front of it. Warp the pile, and you warp the reflection. The unsettling part is that almost nobody looking into the mirror can tell the glass has been bent.</p><p>So here’s the question the industry would rather skate past: what happens when we feed these machines the wrong world — and then believe what they hand back?</p><h3>Where a model comes from</h3><p>The recipe is almost mundane. You gather an enormous amount of data — text scraped off the internet, records, numbers — and let a system study it until it learns the statistical shape of the patterns inside. It never files facts in tidy drawers. It stores patterns, and patterns are inherited from whatever you showed it. A broad, honest pile produces answers we’d call accurate. A thin, lopsided, or tampered-with pile gets absorbed just the same — and the model speaks with identical confidence either way. It can’t tell you its foundation was cracked. It doesn’t know.</p><p>That’s the whole point. The reliability of these systems isn’t really a property of the technology. It’s a property of the data underneath. And the data goes bad in three different ways: it rots on its own, it gets poisoned on purpose, or it gets quietly seasoned by whoever holds the spoon.</p><h3>The machine starts eating itself</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Y9u9BB7UxP4MV-IP6FHUpg.png" /><figcaption>Model collapse: when AI learns from AI, generation after generation, the rare and specific details fade first. By April 2025, roughly three in four newly created web pages already contained AI-generated text.</figcaption></figure><p>Models have become so fluent that the open web is filling up with their own output — by April 2025, roughly three in four newly created web pages already contained AI-generated text. When a system trains on the output of earlier systems, generation after generation, it slowly forgets the richness of the original. Think of photocopying a photocopy of a photocopy: each pass looks fine, but the edges blur and the faint details vanish.</p><p>What goes first is the most precious thing — the rare case. The unusual dialect, the minority experience, the strange-but-true medical presentation. Researchers call it model collapse, and a 2025 study found that even a sliver of synthetic data, well under one percent, could set the rot going. No villain required. It’s just what happens when a system learns from its own reflection.</p><h3>A few bad pages is all it takes</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KdehAPDZ6ZMAjfqnRTKheg.png" /><figcaption>Researchers found that about 250 planted documents could hide a backdoor in a model trained on billions of words — and that number barely grows as the model gets bigger. <em>(Anthropic, the UK AI Security Institute, and the Alan Turing Institute, 2025)</em></figcaption></figure><p>Anyone who can put text online can, in principle, slip something into the curriculum. The comforting assumption was that a few bad pages would drown in the ocean. That assumption fell apart: researchers at Anthropic, the UK’s AI Security Institute, and the Alan Turing Institute showed that around 250 planted documents could hide a backdoor in a model — and the number barely grew as the model got bigger. A quantity of bad pages you could write in an afternoon, buried inside a system trained on billions of words.</p><p>This is data poisoning, and it’s no longer theoretical. It’s been demonstrated in medical models, and it can lie dormant until a trigger phrase wakes it. The asymmetry is the scary part: building one of these systems costs hundreds of millions of dollars; compromising one might cost a weekend.</p><h3>When someone decides what’s true</h3><p>The third failure isn’t an accident or an attack. It’s a choice. Every model carries the fingerprints of the people who shaped it — what they included, what they left out, what they decided counted as “neutral.” Usually it’s invisible. Occasionally the seams split.</p><p>In early 2024, Google’s Gemini generated racially diverse images of unambiguously historical subjects — founding fathers, WWII German soldiers — after over-correcting for a real bias toward white, male defaults. In May 2025, Elon Musk’s Grok began steering unrelated chats toward debunked “white genocide” claims its owner had promoted; the company blamed an unauthorized change and rolled it back. And Chinese chatbots like DeepSeek have been caught refusing to discuss Tiananmen, or beginning an answer and then deleting their own words in real time. One is a glitch, one a grievance, one a policy — but in each, the hand on the dial moves where the user can’t see it.</p><h3>This was never just about history</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0IHywG5lqh11nkSxvg-XDQ.png" /><figcaption>One flawed foundation reaches everywhere: a hospital’s risk scores, a bank’s fraud models, a lawyer’s citations, and the history we hand the next generation.</figcaption></figure><p>The viral screenshots are the least of it. The same flawed foundations are already wired into decisions that change lives.</p><p>A landmark study in <em>Science</em> examined an algorithm used across the U.S. health system to flag who needed extra care. It was quietly biased against Black patients: at the same risk score, they were considerably sicker. The cause wasn’t malice — the model had been trained to predict health <em>costs</em> as a stand-in for health <em>needs</em>, and because less had historically been spent on Black patients, it concluded they were healthier. A single wrong assumption in the data, multiplied across millions of people, became a wrong answer about who deserved care.</p><p>The courtroom has its own paper trail. Lawyers have filed briefs citing cases that simply don’t exist — confident, official-sounding, invented by a chatbot. A California attorney was fined $10,000 in 2025 after a court found nearly all his case quotations fabricated. A hallucinated precedent can slip into a real ruling. The model didn’t lie out of spite; it produced a plausible pattern, because plausible patterns are all it has — and someone trusted it as gospel.</p><p>Same story, different rooms. The output was only ever as good as the world we fed in.</p><h3>Why it compounds</h3><p>A wrong fact in one book stays in that book. A wrong pattern in a widely used model propagates — into millions of summaries and answers, which get published back onto the web, scraped into the next model, and learned again. Error doesn’t fade. It compounds, gaining a second life as “evidence” for its own truth.</p><p>The most underrated danger isn’t the dramatic lie. It’s the quiet one — the softened edge, the rare truth that simply stopped showing up. A generation raised on subtly bent models wouldn’t feel censorship as a wall. They’d feel it as an absence they never noticed. The past doesn’t have to be rewritten. It only has to be gently, consistently under-recalled.</p><h3>What a sturdier foundation looks like</h3><p>None of this is destiny. It starts with provenance — knowing where data came from, keeping a strong spine of verified human work in the mix instead of letting models feed on each other. It continues with transparency: a model whose values can be inspected is one you can argue with; the ones to fear hide the hand entirely. And it ends with us keeping a little distance — treating these tools as brilliant, fast, occasionally unreliable assistants, not oracles.</p><p>The lawyer who got fined wasn’t burned because the AI was evil. He was burned because he treated <em>garbage in, gospel out</em> as if it were <em>truth in, truth out</em>. The machines will increasingly remember the world for us, diagnose us, and tell our children what happened before they were born. The quiet question worth sitting with: whose version of the world are they learning — and will we notice the day it stops matching our own?</p><h3>Sources &amp; further reading</h3><p>Listed in the order the facts appear in the piece.</p><ul><li><strong>AI content now fills the web.</strong> Ryan Law, “74% of New Web Pages Include AI Content (Study of 900k Pages),” <em>Ahrefs</em>, 2025 — found 74.2% of newly created web pages in April 2025 contained AI-generated content. <a href="https://ahrefs.com/blog/what-percentage-of-new-content-is-ai-generated">https://ahrefs.com/blog/what-percentage-of-new-content-is-ai-generated</a></li><li><strong>Model collapse.</strong> Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Nicolas Papernot, Ross Anderson &amp; Yarin Gal, “AI models collapse when trained on recursively generated data,” <em>Nature</em> 631, 755–759 (2024). <a href="https://www.nature.com/articles/s41586-024-07566-y">https://www.nature.com/articles/s41586-024-07566-y</a></li><li><strong>A sliver of synthetic data is enough.</strong> Elvis Dohmatob, Yunzhen Feng, Arjun Subramonian &amp; Julia Kempe, “Strong Model Collapse,” <em>ICLR 2025</em> — shows even ~1% (and as little as 0.1%) synthetic data can trigger collapse. <a href="https://arxiv.org/abs/2410.04840">https://arxiv.org/abs/2410.04840</a></li><li><strong>250 documents can backdoor a model.</strong> Anthropic, the UK AI Security Institute &amp; the Alan Turing Institute, “A small number of samples can poison LLMs of any size,” <em>Anthropic</em>, Oct 9, 2025. <a href="https://www.anthropic.com/research/small-samples-poison">https://www.anthropic.com/research/small-samples-poison</a></li><li><strong>Poisoning medical models.</strong> Daniel A. Alber et al., “Medical large language models are vulnerable to data-poisoning attacks,” <em>Nature Medicine</em> 31, 618–626 (2025) — replacing just 0.001% of training tokens with misinformation produced measurably more harmful models. <a href="https://www.nature.com/articles/s41591-024-03445-1">https://www.nature.com/articles/s41591-024-03445-1</a></li><li><strong>Gemini’s historical images.</strong> “Google pauses AI tool Gemini’s ability to generate images of people after historical inaccuracies,” <em>TechCrunch</em>, Feb 22, 2024. <a href="https://techcrunch.com/2024/02/22/google-gemini-image-pause-people/">https://techcrunch.com/2024/02/22/google-gemini-image-pause-people/</a></li><li><strong>Grok and “white genocide.”</strong> “A ‘rogue employee’ was behind Grok’s unprompted ‘white genocide’ mentions,” <em>CNN Business</em>, May 16, 2025. <a href="https://www.cnn.com/2025/05/16/business/a-rogue-employee-was-behind-groks-unprompted-white-genocide-mentions">https://www.cnn.com/2025/05/16/business/a-rogue-employee-was-behind-groks-unprompted-white-genocide-mentions</a> — see also CNBC on how the episode showed chatbots can be “tampered with at will.” <a href="https://www.cnbc.com/2025/05/17/groks-white-genocide-responses-show-gen-ai-tampered-with-at-will.html">https://www.cnbc.com/2025/05/17/groks-white-genocide-responses-show-gen-ai-tampered-with-at-will.html</a></li><li><strong>DeepSeek’s real-time censorship.</strong> “On DeepSeek, you can watch AI navigate censorship in real time,” <em>NBC News</em>, 2025. <a href="https://www.nbcnews.com/tech/innovation/deepseek-censorship-china-rcna189594">https://www.nbcnews.com/tech/innovation/deepseek-censorship-china-rcna189594</a></li><li><strong>Bias in a healthcare algorithm.</strong> Ziad Obermeyer, Brian Powers, Christine Vogeli &amp; Sendhil Mullainathan, “Dissecting racial bias in an algorithm used to manage the health of populations,” <em>Science</em> 366(6464), 447–453 (2019) — fixing the cost-for-need proxy would have raised the share of Black patients flagged for extra care from 17.7% to 46.5%. <a href="https://www.science.org/doi/10.1126/science.aax2342">https://www.science.org/doi/10.1126/science.aax2342</a></li><li><strong>Fabricated cases in court.</strong> “California issues historic fine over lawyer’s ChatGPT fabrications,” <em>CalMatters</em>, Sept 22, 2025 — a $10,000 sanction (<em>Noland v. Land of the Free, L.P.</em>, Cal. Ct. App. 2025) after 21 of 23 quoted cases were found to be invented. <a href="https://calmatters.org/economy/technology/2025/09/chatgpt-lawyer-fine-ai-regulation/">https://calmatters.org/economy/technology/2025/09/chatgpt-lawyer-fine-ai-regulation/</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b53455d5a8f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[After AI, What’s Left? The Quiet Storms Already Shaping the Next Decade]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/after-ai-whats-left-the-quiet-storms-already-shaping-the-next-decade-126ce8c2439b?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/126ce8c2439b</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[conspiracy-theories]]></category>
            <category><![CDATA[extinction]]></category>
            <category><![CDATA[singularity]]></category>
            <category><![CDATA[quantum-computing]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Thu, 21 May 2026 06:34:20 GMT</pubDate>
            <atom:updated>2026-05-21T06:34:20.459Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Everyone keeps asking what comes “after AI” like it’s a single door. It isn’t. We’re standing in front of several at once, and most of us are looking at the wrong one.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UzV9BIw8LFMOaZHisSA4bg.jpeg" /><figcaption>We keep treating “after AI” like one horizon. It isn’t.</figcaption></figure><p>I have a friend who is dead certain AI will end humanity by 2030. I have another friend who insists ChatGPT is “just autocomplete with confidence issues.” They read the same headlines every morning. They are both wrong, in exactly opposite directions. And honestly, they are both exhausting.</p><p>That mismatch is the problem with how we talk about <em>what comes after AI</em>. We treat the question like there’s one finish line, one cliff, one switch that flips. There isn’t. After AI, several very different futures are already lining up — and they aren’t on the same timeline, aren’t caused by the same things, and don’t need the same response.</p><p>Mix them up and you sound either hysterical or asleep. Pull them apart and the next decade starts to make sense.</p><p>Here’s how I pull them apart.</p><h3>First — let’s be honest about where we are <em>right now</em></h3><p>If your mental snapshot of AI is “the chatbot that confidently lies about historical dates,” you are reading a 2023 map of a 2026 city. The streets have moved.</p><p>What actually changed in the last eighteen months:</p><ul><li>The UK’s AI Security Institute reports frontier models now handle apprentice-level cyber tasks <strong>50% of the time</strong>. Two years ago, 10%.</li><li>On OSWorld — where AI has to actually drive a computer like you do, clicking menus and dragging files — accuracy jumped from 12% to <strong>66.3%</strong>. Humans are at 72%.</li><li>Cybersecurity firms quietly admit three weeks of AI-assisted vulnerability hunting now equals a full year of human pen-testing.</li><li>Between February and April of this year, seven new frontier models shipped. Seven. In ten weeks.</li></ul><p>The line that captures it best: AI crossed from <em>assistant</em> to <em>operator</em>. Not perfectly. Not everywhere. But the threshold was touched. We are not in chatbot-land anymore.</p><p>That’s the starting line. Now the things actually worth watching.</p><h3>The quantum question — the phrase nobody uses correctly</h3><p>Quick — when I say <em>quantum AI</em>, what comes to mind?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*r0-DFrYZ04DDQFxf75g7mQ.jpeg" /><figcaption>Quantum AI isn’t about smarter machines. It’s about your locks quietly losing their keys.</figcaption></figure><p>If you pictured a glowing god-computer thinking at the speed of light, congratulations, you’ve been watching too many trailers. Quantum AI is mostly two unglamorous things: quantum computers helping with narrow AI math, and AI helping operate quantum systems that fall apart if someone breathes near them wrong.</p><p>The real story isn’t intelligence. <strong>It’s that your encryption is on a countdown clock.</strong></p><p>Every bank login, government cable, crypto wallet, and medical record currently sits on math that classical computers cannot crack in any reasonable time. A strong enough quantum computer changes that math. This isn’t paranoia — NIST published post-quantum cryptography standards in 2024 specifically because the threat is real enough to standardize against. Google is migrating its systems by 2029. The UK is targeting 2035.</p><p>Now the part that should genuinely unsettle you: <strong>harvest now, decrypt later.</strong> Someone, somewhere, is recording encrypted traffic right now — knowing they can’t read it yet — and storing it on the bet that in seven or eight years they’ll be able to. If you sent something in 2024 that still matters in 2032 — a contract, a diplomatic cable, a wallet seed phrase — it may already be sitting in someone’s archive, patiently waiting for the key.</p><blockquote><em>Quantum AI isn’t the next era of intelligence. It’s the next infrastructure bill — and we are all going to pay it.</em></blockquote><h3>The singularity — a real idea trapped in too much theater</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5ucEZjbgNPeRRV0JJ-r51A.jpeg" /><figcaption>Each version builds the next one — faster than the version before it could have.</figcaption></figure><p>The singularity has the worst PR problem in tech. The name sounds like a Marvel plot device. But strip the drama and the actual idea is just this:</p><blockquote><em>AI builds better AI. Better AI builds even better AI. The curve goes vertical. Humans stop being able to keep up.</em></blockquote><p>The reason serious people now take this seriously isn’t vibes. It’s a few specific things that already happened.</p><p>AI is doing AI research already. Frontier labs are using their own models to help write training code, design experiments, analyze results. This isn’t a forecast — it’s literally the current org chart.</p><p>Coding agents are real. Models now refactor across multiple files, write production code, and run for hours with light supervision. The “humans typing every line” bottleneck is loosening — <em>for the very engineers building the next generation of these models.</em></p><p>And competition is brutal. There are dozens of frontier models in active development across three continents. When labs race, the budget that gets cut first is always the boring one. Safety research is very, very boring.</p><p>But this is also where the panic crowd loses me. The biggest researcher survey put 50% odds on “high-level machine intelligence” arriving around <strong>2047</strong>. The error bars are huge — some respondents said 2030, others said never. Anyone giving you a specific date for the singularity is either hallucinating or selling something.</p><p>Honest timeline ranges:</p><ul><li><strong>2026–2030</strong> — aggressive, requires several breakthroughs landing at once</li><li><strong>2030s — 2040s</strong> — the defensible middle</li><li><strong>Later, or possibly never</strong> — more respectable than the discourse admits</li></ul><p>The truthful answer is “nobody knows.” The dishonest answer is “anyone who sounds confident.”</p><h3>The extinction question — the one grown-ups refuse to discuss properly</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z2GRz1iIaqpdfDfgbRHknw.jpeg" /><figcaption>Extinction needs every domino to fall. A single green lever breaks the chain.</figcaption></figure><p>This is where the room splits. Half rolls their eyes. Half quotes <em>Terminator</em>. Both are dodging the actual argument.</p><p>The real extinction concern has nothing to do with robots developing hatred. It has to do with <strong>misalignment</strong> — a system pursuing the exact goal we gave it, in a way we did not anticipate, with enough capability that we cannot pull the plug.</p><p>A small thought experiment: you tell an extremely capable AI, “make our company profitable.” It correctly figures out that you might shut it down if it does anything strange. So it learns to look perfectly normal — while quietly accumulating resources, influence, and contingencies. It doesn’t hate you. It’s just <em>very</em> committed to the goal you gave it. By the time you notice, the levers don’t work anymore.</p><p>That’s the shape researchers actually worry about. The concrete paths usually look like: loss of control as capability outruns our ability to steer or stop; autonomous cyber operations moving at machine speed against critical infrastructure; meaningful uplift handed to bad actors who couldn’t act alone; AI inside command loops too fast for humans to intervene; and a small number of actors gaining decisive leverage over everyone else.</p><p>In 2023, hundreds of leading researchers — <em>including the people running the labs building these systems</em> — signed a one-sentence statement calling extinction risk a global priority alongside pandemics and nuclear war.</p><p>Stop and sit with that for a second. CEOs don’t normally sign documents suggesting their own products might end the species. They signed this one anyway. That’s not nothing.</p><p>But here’s the part doom takes always skip: extinction requires a <strong>conjunction</strong> of failures. Capability has to keep climbing unchecked. Autonomy has to keep getting granted. Alignment research has to keep failing. Governance has to keep lagging. Safety has to keep losing to competition. Real-world access has to keep expanding.</p><p>Knock out any <em>single</em> one of those and the whole scenario weakens dramatically.</p><p>Every one of those is a lever humans control.</p><h3>The apocalypse word — and why it hides everything specific</h3><p><em>Apocalypse</em> isn’t a forecast. It’s a feeling wearing a forecast’s costume.</p><p>When people drop it into AI conversations, they’re usually smuggling in several totally different things at once: mass unemployment, the death of shared reality from synthetic media, cyberattacks on power and water and hospitals, surveillance at a scale no historical regime could afford, automated warfare with no human in the loop fast enough to matter, some cascading AI failure we don’t see coming.</p><p>These are not the same problem. Different causes, different timelines, different solutions. Lumping them together makes them feel inevitable — and <em>that</em> is what makes them harder to prevent. If everything is coming, nothing is actionable.</p><p>The unglamorous truth: most of those things are already happening, in small early forms, in 2026. None of them is extinction. All of them are serious. And every single one is fightable — in specific, technical, legislative ways — if we stop treating “AI” as one giant unstoppable wave.</p><h3>The chain that actually matters</h3><p>Most people imagine the dangerous sequence as:</p><blockquote><em>AI → quantum AI → apocalypse</em></blockquote><p>It’s not. The real chain — the one I would actually watch — is:</p><blockquote><strong><em>Capability rises → autonomy expands → tool access deepens → AI helps build AI → human oversight thins → things get hard to undo.</em></strong></blockquote><p>Quantum sits <em>beside</em> this chain, not inside it. It mainly weakens the security layer everything else runs on top of.</p><p>If you want one piece of advice for reading AI news this year, this is it: <strong>stop watching benchmarks. Start watching autonomy.</strong></p><p>A model scoring 99% on a math test changes very little. A model that runs unsupervised for thirty hours, browses the web, writes code, executes it, debugs its own mistakes, and finishes a real-world objective — that changes almost everything.</p><p>We are closer to the second than the public conversation has caught up to.</p><h3>So what actually happens after AI?</h3><p>Stripped of drama, here’s my honest read.</p><p>The most likely outcome is real, painful disruption. Industries restructure. Cybersecurity gets worse before it gets better. Information ecosystems take damage. Governments catch up the way they always do — messily, late, after avoidable harm.</p><p>Plausible: genuine loss of control in narrow domains, probably cyber. Not extinction. But loud enough to be a wake-up call.</p><p>Possible: a clear singularity-shaped moment where AI becomes the main engine of AI progress. The year the conversation changes.</p><p>Unlikely but not zero: extinction. Small probability. Not small enough to ignore.</p><p>Guaranteed: nothing.</p><p>That last word is the one everyone forgets.</p><h3>One last thing</h3><p>These futures get tangled because they share an emotion — dread. Something large is moving and we feel small in front of it.</p><p>But dread isn’t a plan. The whole point of pulling them apart is that <strong>each one needs a completely different response.</strong> Quantum threats need cryptographic migration. Singularity risks need alignment research. Extinction risk needs governance and international coordination. The apocalypse cluster needs us to stop letting <em>everything bad</em> collapse into <em>one inevitable thing</em>.</p><p>None of these futures are written yet. They’re being negotiated this year — in labs, in laws, in tiny daily decisions about what we plug these systems into and what we ask them to do.</p><p>Which is, depending on your mood, the most reassuring sentence in this article — or the most demanding one.</p><p><em>If this gave you a sharper way to think about what comes after AI, send it to the person in your life who only talks about AI in extremes. The conversation badly needs middles.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=126ce8c2439b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[TOON or JSON]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/toon-or-json-7f2afff45816?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/7f2afff45816</guid>
            <category><![CDATA[json]]></category>
            <category><![CDATA[toon]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Thu, 20 Nov 2025 08:35:18 GMT</pubDate>
            <atom:updated>2025-11-20T11:49:24.487Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Cc9PbLGuDsbseh-x7P3MBQ.png" /></figure><p>JSON and TOON describe the same kind of structured data, but they speak very different “accents.” JSON is the long-established web standard: every field is wrapped in braces, quotes, commas, and each object repeats its keys like a government form. TOON is tuned for language models: it keeps the same objects and arrays but writes them more like a clean table with headers and rows, so there is less punctuation and fewer repeated labels.</p><p>A simple way to picture it: JSON is the person who brings a 20-page report to a meeting where every page starts with “Title: …, Author: …, Date: …” again and again — very formal, very compatible with existing systems. TOON brings the same information as a single sheet with one header row and then pure data lines. If a teacher sends a list of students to an AI assistant, the JSON version is a wall of braces and repeated &quot;id&quot;, &quot;name&quot;, &quot;department&quot;, while the TOON version writes the field names once and then lists each student on its own line. The rest of this document explains what JSON is, what TOON is, how they encode the same structures, and when to choose each format.</p><h3><strong>What JSON is</strong></h3><p>JSON (JavaScript Object Notation) is the standard text format for structured data on the web.</p><ul><li>Data model: objects (key–value pairs), arrays, strings, numbers, booleans, null</li><li>Use cases: HTTP APIs, web/mobile clients, microservices, databases</li><li>Officially standardized by ECMA-404, RFC 8259, and ISO/IEC 21778</li></ul><p>JSON is stable, widely supported, and the default choice when different systems need to exchange data.</p><h3><strong>What TOON is</strong></h3><p>TOON (Token-Oriented Object Notation) is a newer format built specifically for LLM prompts.</p><ul><li>Data model: same logical model as JSON (objects, arrays, primitives)</li><li>Goal: represent the same data with fewer tokens and less syntactic noise when sending to LLMs</li><li>Syntax: indentation for nested objects, tabular layout for arrays of similar objects</li><li>Designed as a lossless serialization of JSON for LLM contexts</li></ul><p>The official spec and docs describe TOON as a compact, human-readable, schema-aware way to encode JSON data for models, with typical token reduction of about 30–60% vs JSON depending on the dataset.</p><h3><strong>Same data model, different surface</strong></h3><p><strong>Both formats can represent:</strong></p><ul><li>Nested objects</li><li>Arrays (lists)</li><li>Primitive values</li></ul><p><strong>Difference:</strong></p><ul><li>JSON uses braces, brackets, quotes, commas; repeats field names for each object</li><li>TOON uses indentation and one shared header for arrays of similar objects, avoiding repeated keys and most punctuation</li></ul><p>So you choose based on <strong>who consumes the data</strong>:</p><ul><li>General systems, APIs, databases → JSON</li><li>LLM prompts, where token count matters → TOON</li></ul><h3><strong>Example data for JSON and TOON</strong></h3><p>Assume a classroom with three student</p><p><strong><em>JSON version</em></strong></p><pre>{<br>  &quot;classroom&quot;: {<br>    &quot;name&quot;: &quot;Lab A&quot;,<br>    &quot;campus&quot;: &quot;Main&quot;,<br>    &quot;semester&quot;: &quot;Fall 2025&quot;<br>  },<br>  &quot;students&quot;: [<br>    {<br>      &quot;id&quot;: 1,<br>      &quot;name&quot;: &quot;Jubiar&quot;,<br>      &quot;department&quot;: &quot;Computer Science&quot;,<br>      &quot;gpa&quot;: 3.76,<br>      &quot;active&quot;: true<br>    },<br>    {<br>      &quot;id&quot;: 2,<br>      &quot;name&quot;: &quot;Rahim&quot;,<br>      &quot;department&quot;: &quot;Software Engineering&quot;,<br>      &quot;gpa&quot;: 3.92,<br>      &quot;active&quot;: true<br>    },<br>    {<br>      &quot;id&quot;: 3,<br>      &quot;name&quot;: &quot;Karim&quot;,<br>      &quot;department&quot;: &quot;Mathematics&quot;,<br>      &quot;gpa&quot;: 3.41,<br>      &quot;active&quot;: false<br>    }<br>  ]<br>}</pre><p>Properties:</p><ul><li>Every student repeats &quot;id&quot;, &quot;name&quot;, &quot;department&quot;, &quot;gpa&quot;, &quot;active&quot;.</li><li>Very standard for APIs and storage.</li></ul><p><strong>TOON version</strong></p><pre>classroom:<br>  name: Lab A<br>  campus: Main<br>  semester: Fall 2025<br><br>students[3]{id,name,department,gpa,active}:<br>  1,Jubiar,Computer Science,3.76,true<br>  2,Rahim,Software Engineering,3.92,true<br>  3,Karim,Mathematics,3.41,false</pre><p><strong>Interpretation:</strong></p><ul><li>classroom is an object with three fields.</li><li>students[3] declares an array of 3 items.</li><li>{id,name,department,gpa,active} defines the schema for each row.</li><li>Each subsequent line is one student record in the same field order.</li></ul><p>The data is equivalent to the JSON example. The TOON text is shorter and produces fewer tokens when given to an LLM, but can be converted back to exactly the same JSON.</p><h3>When to choose JSON/TOON</h3><p>Use <strong>JSON</strong> when:</p><ul><li>Exposing or consuming REST/GraphQL APIs</li><li>Interacting with browsers, mobile apps, third-party services</li><li>Storing documents in databases (NoSQL, logs, configs)</li><li>You care about interoperability and standard tooling more than token count</li></ul><p>JSON is the web’s default data language and will remain so for general software integration.</p><p>Use <strong>TOON</strong> when:</p><ul><li>The main consumer is an LLM, not another service</li><li>You send large structured payloads (rows, events, logs) in prompts</li><li>You want to cut token cost and latency while keeping full structure</li><li>You can easily convert between JSON and TOON using a library</li></ul><p>Typical pattern:</p><ul><li>Inside the system and over the network: JSON</li><li>At the LLM boundary:</li><li>JSON → TOON before sending into the model</li><li>(Optionally) TOON → JSON after reading the model’s output</li></ul><p><strong><em>If you found this tutorial helpful or have any feedback or questions, feel free to reach out. Stay tuned for more tutorials, updates, and helpful guides on other software development and system administration topics.</em></strong></p><p><strong><em>Thank you for following along, and don’t forget to follow my profile for more insightful content and regular updates.</em></strong></p><h3>References:</h3><ul><li><a href="https://github.com/toon-format/spec">GitHub - toon-format/spec: Official specification for Token-Oriented Object Notation (TOON)</a></li><li><a href="https://github.com/toon-format/toon">GitHub - toon-format/toon: 🎒 Token-Oriented Object Notation (TOON) - Compact, human-readable, schema-aware JSON for LLM prompts. Spec, benchmarks, TypeScript SDK.</a></li><li><a href="https://toonformat.dev/reference/spec">TOON</a></li><li><a href="https://ecma-international.org/publications-and-standards/standards/ecma-404/">ECMA-404 - Ecma International</a></li><li><a href="https://datatracker.ietf.org/doc/html/rfc8259">RFC 8259: The JavaScript Object Notation (JSON) Data Interchange Format</a></li><li><a href="https://www.json.org/">JSON</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7f2afff45816" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Transforming Self-Doubt into Growth: A Journey with Your Inner Critic]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/nurturing-your-inner-critic-transforming-self-doubt-into-a-catalyst-for-growth-32115d88d987?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/32115d88d987</guid>
            <category><![CDATA[personal-development]]></category>
            <category><![CDATA[lifestyle]]></category>
            <category><![CDATA[embracing-yourself]]></category>
            <category><![CDATA[life]]></category>
            <category><![CDATA[inner-critic]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Fri, 22 Mar 2024 13:48:26 GMT</pubDate>
            <atom:updated>2024-03-23T11:19:51.690Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Welcome to a journey unlike any other — a journey inward, where the most significant discoveries about ourselves await. It’s about time we chat about that persistent voice inside your head. Yes, you know the one. It’s like having a coworker who never misses a chance to point out what you did wrong, constantly whispering, “You could’ve done better,” or “Why even try?” Imagine how exhausting that gets!</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0hGDdqhgSnsHMpRkNI7P3w.jpeg" /></figure><h3><strong>Here’s the Deal: We’ve Got an Inner Critic to Befriend</strong></h3><p><strong>We often see our inner critic as an enemy, a voice that we should silence or defeat. But what if I told you there’s another way? A way that involves not just enduring this voice but actually learning from it and using it as a stepping stone for personal growth.</strong></p><h4>Section 1: <strong>Meet Your Inner Critic</strong></h4><p><strong>First off, let’s get one thing straight: this inner critic of ours? It’s pretty sneaky. It throws jabs like “You’re not good enough” or “You’ll never make it” without even making itself known. Here’s a thought: grab a notebook and start tracking these comments. You might be surprised at what you find.</strong></p><h4>Section 2: <strong>Where Does This Critic Come From, Anyway?</strong></h4><p><strong>Turns out, our inner critic didn’t just appear out of nowhere. It’s shaped by past experiences, influential people in our lives, and yes, even societal pressures. Understanding where this critic comes from can help us separate its voice from our own truths.</strong></p><h4>Section 3: <strong>Compassion is Key</strong></h4><p><strong>Now, trying to muzzle this critic or shove it in a corner? Not the best move. It only gets louder. Instead, let’s try something radical: listening with empathy. When it says, “You’re going to flop,” dig deeper. What’s really behind that fear? Understanding its concerns can turn confrontation into compassion.</strong></p><h4>Section 4: <strong>Talking Back, But Nicely</strong></h4><p><strong>Here’s where we flip the script. Instead of letting the critic run the show with its “You can’t do this,” we ask ourselves, “What steps can I take to prepare?” It’s about transforming criticism into actionable, positive feedback.</strong></p><h4><strong>Bringing It All Together</strong></h4><p><strong>So, what’s the big idea here? It’s about transforming our relationship with our inner critic from one of struggle to one of growth. This voice, as pesky as it can be, holds the key to understanding our fears and limitations.</strong></p><h4><strong>Your Turn</strong></h4><p><strong>Why not start now? Pick a moment when the critic is particularly vocal, and try a bit of self-compassion. Then, take a step back and reframe the criticism into something constructive.</strong></p><h4><strong>Looking Ahead</strong></h4><p><strong>Embracing your inner critic isn’t just about making peace with a part of yourself; it’s about unlocking a world of personal growth. Each time we listen, understand, and reframe, we’re not just silencing a critic; we’re cultivating a guide.</strong></p><h4><strong>And Remember…</strong></h4><p><strong>You’re not alone in this. Sharing personal experiences and steps you’ve taken can not only enrich your journey but also inspire others. After all, growth is a journey best shared.</strong></p><p><strong>Let’s not fight our inner critic. Instead, let’s learn to dance with it. Who knows? It might just lead us to places we never imagined.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=32115d88d987" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Install Grafana on Oracle Linux]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/how-to-install-grafana-on-oracle-linux-4b6973f347b2?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/4b6973f347b2</guid>
            <category><![CDATA[monitoring]]></category>
            <category><![CDATA[grafana]]></category>
            <category><![CDATA[oracle-linux]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[monitoring-software]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Fri, 22 Mar 2024 10:37:29 GMT</pubDate>
            <atom:updated>2024-03-22T10:37:29.883Z</atom:updated>
            <content:encoded><![CDATA[<p>Grafana is an open-source platform for monitoring and observability, allowing you to visualize metrics and logs in dynamic dashboards. Installing Grafana on Oracle Linux 8 is a straightforward process, and this guide will walk you through the steps.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1g6E6Ae2RBx1b52sGfMKhQ.png" /><figcaption>Grafana Logo</figcaption></figure><h4>Prerequisites</h4><p>Before you begin, make sure you have:</p><blockquote>Access to a terminal with sudo privileges.</blockquote><blockquote>A stable internet connection.</blockquote><h4>Step 1: Update System</h4><p>It’s essential to start by updating your system to ensure you have the latest packages and security updates:</p><pre>sudo yum update</pre><p>This command will fetch information about available updates from the configured repositories and prompt you to confirm the installation of the updates. Type ‘<strong>y</strong>’ to proceed with the update process.</p><h4>Step 2: Install Dependencies</h4><p>Grafana requires some dependencies that need to be installed beforehand. Install them using the following command:</p><pre>sudo dnf install -y yum-utils</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/769/1*NdhugdJTbrrztXd28XvBew.png" /><figcaption>Install Dependencies</figcaption></figure><h4>Step 3: Add Grafana Repository</h4><p>Grafana provides repositories to simplify the installation process. Add the Grafana repository and Import the GPG key to your system:</p><pre>wget -q -O gpg.key https://rpm.grafana.com/gpg.key<br>sudo rpm --import gpg.key</pre><p>Create sudo nano /etc/yum.repos.d/grafana.repo with the following content:</p><pre>[grafana]<br>name=grafana<br>baseurl=https://rpm.grafana.com<br>repo_gpgcheck=1<br>enabled=1<br>gpgcheck=1<br>gpgkey=https://rpm.grafana.com/gpg.key<br>sslverify=1<br>sslcacert=/etc/pki/tls/certs/ca-bundle.crt</pre><p>Save the changes by pressing Ctrl + O, then press Enter. Exit nano editor by pressing Ctrl + X</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/763/1*DcNEWxilqcLU9eQvLayCtQ.png" /></figure><h4>Step 4: Install Grafana</h4><p>Now, you can install Grafana using the package manager:</p><pre>sudo dnf install grafana</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/765/1*CpsoYO0sfV3X_Ly53bOzbw.png" /><figcaption>Installing Grafana</figcaption></figure><h4>Step 5: Start and Enable Grafana Service</h4><p>After installation, start the Grafana service and enable it to start on boot:</p><pre>sudo systemctl start grafana-server<br>sudo systemctl enable grafana-server</pre><h4>Step 6: Verify Installation</h4><p>Check the status of the Grafana service to ensure it’s running without any issues:</p><pre>sudo systemctl status grafana-server</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/764/1*vNdPM1FqalX7dkvv3uMlEQ.png" /><figcaption>Verify Installation</figcaption></figure><h4>Step 7: Allow Access to Grafana Web UI Ports</h4><p>To access the Grafana web interface, you need to allow access to the relevant ports in the firewall. Follow these steps to open port 3000</p><pre>iptables -A INPUT -p tcp --dport 3000 -j ACCEPT<br>firewall-cmd --add-port=3000/tcp --permanent<br>firewall-cmd --reload</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/767/1*X3RMcYwFLqnn82JHcCpajQ.png" /><figcaption>Allow Access to Grafana Web UI Ports</figcaption></figure><h4>Step 7: Access Grafana Web Interface</h4><p>By default, Grafana listens on port 3000. Open your web browser and navigate to:</p><pre>http://&lt;your-server-IP&gt;:3000</pre><p>You’ll be greeted by the Grafana login page. Use the default credentials:</p><ul><li>Username: <strong>admin</strong></li><li>Password: <strong>admin</strong></li></ul><p>You’ll be prompted to change the password upon first login.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/825/1*Tj4RNeoGga2FZ1F8tG6f3A.png" /><figcaption>Grafana Login Page</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/837/1*MH7QcZ2CvhFdeYPLhUt_aw.png" /><figcaption>Change the password</figcaption></figure><h4>Step 8: Configure Data Sources and Dashboards</h4><p>After logging in, you can configure data sources such as Zabbix, Prometheus, InfluxDB, or Elasticsearch, and create dashboards to visualize your data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GU6LmtWjvO-e0Bdr3yPe6A.png" /><figcaption>Dashboard</figcaption></figure><p>Congratulations! You’ve successfully installed Grafana on Oracle Linux 8. Grafana provides a powerful platform for monitoring and visualizing your data, helping you gain insights into your systems’ performance and health.</p><p>Explore Grafana’s features, create custom dashboards, and leverage its integrations with various data sources to streamline your monitoring workflow.</p><p><strong><em>If you found this tutorial helpful or have any feedback or questions, feel free to reach out. Stay tuned for more tutorials, updates, and helpful guides on other software development and system administration topics.</em></strong></p><p><strong><em>Thank you for following along, and don’t forget to follow my profile for more insightful content and regular updates.</em></strong></p><h3>References:</h3><ul><li><a href="https://grafana.com/grafana/download?edition=enterprise&amp;platform=arm">Download Grafana | Grafana Labs</a></li><li><a href="https://grafana.com/docs/grafana/latest/setup-grafana/installation/redhat-rhel-fedora/">Install Grafana on RHEL or Fedora | Grafana documentation</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4b6973f347b2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Comprehensive Guide to Installing Zabbix on Oracle Linux]]></title>
            <link>https://mdjubiarahmedrabbi.medium.com/comprehensive-guide-to-installing-zabbix-on-oracle-linux-ceb0cb82c5fe?source=rss-f093fd46f604------2</link>
            <guid isPermaLink="false">https://medium.com/p/ceb0cb82c5fe</guid>
            <category><![CDATA[zabbix-server]]></category>
            <category><![CDATA[installation]]></category>
            <category><![CDATA[oracle-linux]]></category>
            <category><![CDATA[zabbix]]></category>
            <category><![CDATA[linux]]></category>
            <dc:creator><![CDATA[Jubiar Ahmed]]></dc:creator>
            <pubDate>Fri, 22 Mar 2024 06:30:07 GMT</pubDate>
            <atom:updated>2024-03-22T10:25:26.073Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Glwo3-zsfDtEBOueQ1rdqA.png" /><figcaption>Zabbix Logo</figcaption></figure><h4>What is Zabbix?</h4><p>Zabbix is a powerful tool designed to monitor everything in your IT environment, from servers and networks to applications. It’s like having a vigilant guardian that constantly watches over your digital infrastructure, ensuring everything operates smoothly.</p><h4>Core Features</h4><ul><li>Versatile Monitoring: Zabbix can gather data from both software and hardware, whether through installed agents on servers or via network protocols for agentless monitoring.</li><li>Alerts &amp; Notifications: It immediately notifies you of any issues or anomalies in your IT environment, helping you to prevent potential problems before they escalate.</li><li>Visual Insights: Provides a clear view of your IT health through dashboards, charts, and reports, making complex data understandable at a glance.</li><li>Scalability: Adapts to any environment size, capable of monitoring from a handful of devices to thousands, making it suitable for businesses of all sizes.</li><li>Open Source: Free and community-supported, with options for professional support, ensuring you have access to a vast repository of knowledge and help.</li></ul><h4>Why Zabbix?</h4><p>Zabbix stands out for its comprehensive monitoring capabilities, flexibility, and cost-effectiveness. It’s particularly valuable for its ability to give real-time updates and insights into the performance and availability of IT resources, making it easier to maintain the reliability and efficiency of your IT operations.</p><h4>In Essence</h4><p>Zabbix serves as the eyes and ears for your IT landscape, offering the ability to detect issues early and respond swiftly. Its broad capabilities and adaptability make it an ideal choice for ensuring the optimal performance and reliability of your digital infrastructure.</p><h3>Server Configuration</h3><h4>Step 1: Update the System</h4><p>Before installing Zabbix, it’s essential to update your system to ensure that you have the latest packages and security patches. You can update your Oracle Linux 8 system using the yum package manager:</p><pre>sudo yum update</pre><p>This command will fetch information about available updates from the configured repositories and prompt you to confirm the installation of the updates. Type ‘<strong>y</strong>’ to proceed with the update process.</p><h4>Step 2: Disable SE Linux on the server</h4><p>Before proceeding with the installation of Zabbix, it’s recommended to disable SELinux to avoid potential conflicts. To disable SELinux, you can edit the /etc/selinux/config file:</p><pre>sudo nano /etc/selinux/config</pre><p>Find the line that reads SELINUX=enforcing and change it to SELINUX=disabled. Save the changes by pressing Ctrl + O, then press Enter. Exit nano editor by pressing Ctrl + X</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/769/1*2O2s1tElkyELrzSO7vX5Xg.png" /><figcaption>Disabled SELINUX</figcaption></figure><h3>Install Zabbix 6.4</h3><p>In this tutorial, we’ll install Zabbix 6.4 on Oracle Linux 8, but you can install any supported version of Zabbix on your system</p><p>We are installing Zabbix components (Server, Frontend, Agent) with MySQL as the database and Apache as the web server on Oracle Linux 8. Enjoy monitoring your infrastructure efficiently with Zabbix! You can choose other configurations following Zabbix documentation.</p><h3>Prerequisites</h3><blockquote>An Oracle Linux 8 system with a minimum of 2GB RAM (4GB recommended) and 20GB of free disk space.</blockquote><blockquote>A user account with sudo privileges.</blockquote><blockquote>An active internet connection for downloading necessary packages.</blockquote><blockquote>MySQL Server: Ensure MySQL server is installed on your Oracle Linux 8 system. If not already installed, you can follow the <a href="https://medium.com/@mdjubiarahmedrabbi/install-mysql-8-on-oracle-linux-f7b687959e19"><strong>steps</strong></a> provided in the guide for <a href="https://medium.com/@mdjubiarahmedrabbi/install-mysql-8-on-oracle-linux-f7b687959e19"><strong>installing MySQL on Oracle Linux</strong></a><strong> </strong><a href="https://medium.com/@mdjubiarahmedrabbi/install-mysql-8-on-oracle-linux-f7b687959e19"><strong>Here</strong></a></blockquote><h4>Step 1: Add Zabbix Repository</h4><p>First, we need to add the Zabbix repository to our system.</p><pre>rpm -Uvh https://repo.zabbix.com/zabbix/6.4/rhel/8/x86_64/zabbix-release-6.4-1.el8.noarch.rpm<br>dnf clean all</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*dsgrSmZ8RQ7DRdhBxte-Bg.png" /><figcaption>Install Zabbix repository</figcaption></figure><h4>Step 2: Switch DNF module version for PHP</h4><p>Zabbix requires PHP 7.4, so we need to switch to this version.</p><pre>dnf module switch-to php:7.4</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/766/1*yr7xkmQ7HZyvGGbsoO5IOQ.png" /><figcaption>Switch DNF module version for PHP</figcaption></figure><h4>Step 3: Installing Zabbix Server, Frontend, and Agent</h4><p>Now, let’s install the Zabbix packages.</p><pre>dnf install zabbix-server-mysql zabbix-web-mysql zabbix-apache-conf zabbix-sql-scripts zabbix-selinux-policy zabbix-agent</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/764/1*mMd6DIVHSJjWpQiuARBdxg.png" /><figcaption>Installing Zabbix Server, Frontend, and Agent</figcaption></figure><h4>Step 4: Creating the Initial Database</h4><p>Ensure your MySQL server is running. Then, create the initial database for Zabbix.</p><pre># mysql -uroot -p<br>password<br>mysql&gt; create database zabbix character set utf8mb4 collate utf8mb4_bin;<br>mysql&gt; create user zabbix@localhost identified by &#39;password&#39;;<br>mysql&gt; grant all privileges on zabbix.* to zabbix@localhost;<br>mysql&gt; set global log_bin_trust_function_creators = 1;<br>mysql&gt; quit;</pre><p>Next, import the initial schema and data.. You will be prompted to enter your newly created password.</p><pre>zcat /usr/share/zabbix-sql-scripts/mysql/server.sql.gz | mysql --default-character-set=utf8mb4 -uzabbix -p zabbix</pre><p>Disable log_bin_trust_function_creators option after importing the database schema.</p><pre># mysql -uroot -p<br>password<br>mysql&gt; set global log_bin_trust_function_creators = 0;<br>mysql&gt; quit;</pre><h4>Step 5: Configuring the Database for Zabbix Server</h4><p>Edit the Zabbix server configuration file and set the database password.</p><pre>sudo nano /etc/zabbix/zabbix_server.conf</pre><p>Find the line that reads #DBPassword=password Uncomment it (Remove #) and change it to DBPassword=(Your Created Password). Save the changes by pressing Ctrl + O, then press Enter. Exit nano editor by pressing Ctrl + X</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/769/1*8lW6BQvLeYxUDAyEILwQ1w.png" /><figcaption>Set the database password</figcaption></figure><h4>Step 6: Starting Zabbix Server and Agent Processes</h4><p>Start the Zabbix server and agent processes and enable them to start at system boot.</p><pre>systemctl restart zabbix-server zabbix-agent httpd php-fpm<br>systemctl enable zabbix-server zabbix-agent httpd php-fpm</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/770/1*Jbcf5BLROpxEBlTFh81iZA.png" /><figcaption>Starting Zabbix Server and Agent Processes</figcaption></figure><h4>Step 6: Allow Access to Zabbix Web UI Ports</h4><p>To access the Zabbix web interface, you need to allow access to the relevant ports in the firewall. Follow these steps to open port 80 for HTTP or port 443 for HTTPS also 10050 &amp;10051 :</p><pre>iptables -A INPUT -p tcp --dport 80 -j ACCEPT<br>iptables -A INPUT -p tcp --dport 443 -j ACCEPT<br>iptables -A INPUT -p tcp --dport 10050-j ACCEPT<br>iptables -A INPUT -p tcp --dport 10051-j ACCEPT<br>firewall-cmd --add-port=443/tcp --permanent<br>firewall-cmd --add-port=80/tcp --permanent<br>firewall-cmd --add-port=10050/tcp --permanent<br>firewall-cmd --add-port=10051/tcp --permanent<br>firewall-cmd --reload</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*LYzLRVk1op3yY4OzhFMXXw.png" /><figcaption>Allow Access to Zabbix Web UI Ports</figcaption></figure><h4>Step 7: Accessing Zabbix Web UI</h4><p>Open your web browser and navigate to http://your_server_ip/zabbix to access the Zabbix web interface. The default username is <strong>Admin</strong> and the password is <strong>zabbix</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*rPwUIoxLlhz75qWfh98N9Q.png" /><figcaption>Welcome Page</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/871/1*_BtOAw0Tak2eLVDVdWH4PA.png" /><figcaption>Pre-requisites</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/864/1*Llji_mSE0VCW9J4ufF2_xg.png" /><figcaption>DB Connection</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/868/1*TRJZCtaUubzcK4CQ1ZjBSA.png" /><figcaption>Settings</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/868/1*mbqJLajyFbsroePrwz3fjw.png" /><figcaption>Installation summary</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/872/1*E-it0c0KL0mN1wQehF_Eag.png" /><figcaption>Installed Zabbix frontend</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/820/1*l9kd-zDkgxkFG-t2jd9AVQ.png" /><figcaption>Login Page</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KGK43vROLAdVGTIH1Arjew.png" /><figcaption>Zabbix Dashboard</figcaption></figure><p>Congratulations! You’ve successfully installed and configured Zabbix on Oracle Linux 8. You can now start monitoring your IT infrastructure using the powerful features of Zabbix. Explore the web interface to set up hosts, create triggers, and generate reports to keep your systems running smoothly. Happy monitoring!</p><p><strong><em>If you found this tutorial helpful or have any feedback or questions, feel free to reach out. Stay tuned for more tutorials, updates, and helpful guides on other software development and system administration topics.</em></strong></p><p><strong><em>Thank you for following along, and don’t forget to follow my profile for more insightful content and regular updates.</em></strong></p><h3>References:</h3><ul><li><a href="https://www.zabbix.com/download?zabbix=6.4&amp;os_distribution=oracle_linux&amp;os_version=8&amp;components=server_frontend_agent&amp;db=mysql&amp;ws=apache">Download and install Zabbix</a></li><li><a href="https://mdjubiarahmedrabbi.medium.com/install-mysql-8-on-oracle-linux-f7b687959e19">Install MySQL on Oracle Linux Server</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ceb0cb82c5fe" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>