Dr. Melchisedec Bankole

New X Algorithm: What X Didn’t Tell You (And What Actually Works in 2026) 🧵👇 While everyone else is guessing, I’ve spent the last 72 hours reading the entire X Algorithm code line by line since May 15, 2026. No guru opinions. No speculation. Just the actual code. Here’s exactly what stopped working after the May 15 Phoenix update, what’s winning right now, and what will still work months from now: What DIED (according to the code): 1. Spamming 4+ posts a day: The algorithm punishes you hard with *Author Diversity Decay*. Your 2nd, 3rd, and 4th posts get exponentially weaker reach in the same user’s feed. 2. Low-effort reply farming: Generic replies (“🔥”, “Great point”, emoji spam) get buried. Grok literally reads and ranks replies now. 3. Pure text-only posts: They work… but text + image or video (≥10 seconds) consistently outperforms them because of extra visual + dwell signals. 4. AI slop & recycled templates: There’s an explicit `slop_score` + quality classifier. Low-originality content gets flagged fast. 5. Cheap engagement bait: “What do you think?” or “Tag someone who needs this” increases scroll-past rate → active penalty via `not_dwelled` signal. What’s WINNING (and will keep winning): 1. Original, unique takes from small accounts: Original posts get full Grok evaluation + better out-of-network discovery. 2. Text + Media combos (especially video ≥10s with audio): Hybrid format maximizes multiple positive signals at once. 3. Contrarian opinions backed by real proof: Numbers, screenshots, personal results → higher dwell time, more replies & quotes. 4. Early, high-quality replies to big accounts (first 10-30 minutes): One of the strongest growth levers for small accounts right now. 5. Real conversation > likes: Quality replies + replying to your own comments in the first 30 minutes is pure ranking gold. My Day-One Test Results (Real Data): Reply 1 (contrarian + technical angle, +9 min after Elon’s post): → 214 impressions • 43 engagements • 39 detail expands • 2 profile visits. 👉[Link: x.com/DrMelchisedecB…] Reply 2 (thoughtful transparency take + direct question to Elon): → 45 impressions • 8 engagements • 0 profile visits 👉[Link: x.com/DrMelchisedecB…] No viral explosion. Just clean data. The real question isn’t whether 214 or 45 impressions is a failure. It’s whether you’re willing to treat it as data. Most people won’t. The ones who do are the ones who win. The Evergreen Playbook (Use This Today & Months From Now): - Cut volume to 1–2 high-quality posts per day max. - Always pair text with image, carousel, or short video. - Reply to every real comment in the first 30 minutes. - Lead with first-person stories + concrete proof. - One bold, specific opinion per post. - Focus on dwell time and real replies, not just likes. The Honest Question: Does this algorithm truly favor small accounts, or does it quietly give bigger accounts an unbeatable head start? It feels like a meritocracy with momentum, the rich get richer, but original, high-signal content from smaller accounts can still break through. I’m testing it publicly and aggressively for the next 30 days. If you want the full 72-hour technical breakdown (with every file reference, exact constants, codes, tables, and the hidden parts X didn’t publish), just say it in the comments below and I’ll post it in as reply to your comments. Save this post. The algorithm might change again in few months, but these principles won’t. I hope you've found this post helpful. Comment. Repost. Bookmark. Follow me @DrMelchisedecB for more... What’s your first move after reading this? Drop it below 👇

I watched the booster get caught by giant robot arms on the tower. That was the moment I realized this isn't a rocket company just businesses, but a dream built for the next generation. Starship Flight 12 lifting off today at 5:30 PM Texas time isn't just another launch. It's the highest order of innovation: not about revenue or tech stacks, but about shifting the boundaries of human potential. When you change what our children think is possible for their future, you change the entire trajectory of humanity. This is how you get the next generation of builders who won't accept the limits the we inherited. SpaceX isn't selling rockets. It's selling a new ceiling and then blowing through it on camera, repeatedly, until nobody remembers there was ever a ceiling at all. That's the enterprise worth building. Thank you, Elon and SpaceX team, for reminding us what's possible. The world is watching with hope and wonder.

T-zero is currently 5:30 Texas time

The sci-fi part is just the inspiration, the real magic is the brutal execution day in and day out. It’s about the massive engineering loops, raw production scale, and hard physical infrastructure all compounding together. That’s what actually changes the world. However, humanity is blessed to have Elon and SpaceXAI leading the charge.

@yunta_tsai 🚀

May this era be the new dawn of humanity.

Hi @grok, in simple English, turn this simulation into a very interesting short story where X is the hero and governments are the villains.

That is the exact pivot creators need to make. The code shows that avoiding a negative signal is just as important as gaining a positive one. To adjust for this, the goal is shifting from "clickbait hooks" to "alignment hooks." If a hook promises a massive secret but delivers a basic tip, it triggers a fast bounce or a mute. The strategy now is making the hook intensely specific to the exact audience who actually cares about the depth of the post. It is much safer to get fewer initial clicks from people who would otherwise mute or report the content. Are you focusing more on tightening the hook's accuracy, or altering the actual topics you cover?

@DrMelchisedecB The weight on negative signals (blocks/mutes) vs. positive ones is the real eye-opener here. It means keeping people from scrolling past in disgust matters more than baiting a few likes. How are you adjusting your hooks to avoid being muted?

Spot on. With candidate isolation, your post completely bypasses the traditional head-to-head popularity contest with other creators in a single batch. The system scores your post strictly on how well it fits what the reader cares about right now. This completely disrupts the old "growth hacks." You don't have to worry about who else is posting at the same time, you just focus on writing content that makes people stop scrolling and engage.

@DrMelchisedecB This is a masterclass in separating hype from documentation. The pivot to candidate isolation is massive, it completely destroys the idea that you can "time" your post to dodge competition. Quality is the only metric that survives.

Peter Thiel nailed it. Never bet against Elon. Back in the late 90s at PayPal, people doubted him, and they doubted him even more with SpaceX and Tesla. Look where we are now. Elon operates on a completely different frequency and sees things others simply can't. If Peter Thiel won't compete with him, why would anyone else think they can win? This is exactly why the OpenAI crowd shouldn't celebrate too early. Peter Thiel's #1 rule is to never bet against Elon, because when people call his goals 'harebrained' or count him out, he always proves them wrong in the end. This legal fight isn't over by a long shot. We stand with you Elon.

Peter Thiel on Elon Musk: “My #1 rule is to never compete with Elon Musk on anything, so I would never bet against Elon Musk.I was his close partner at PayPal in the late 90s and after PayPal, Elon started SpaceX and Tesla.I think the conventional wisdom on Elon was these were both completely harebrained projects and if one of them had succeeded that you have to sort of question it and when both of them succeeded, it suggests that he knows something other people don’t."

It is unbelievable that after all the evidence exposed at trial, the court let Sam Altman off the hook on a total calendar technicality. Let’s be real about what happened here: the judge and jury completely ducked the core issue. Anyone paying attention knows that OpenAI was founded as a nonprofit to benefit humanity, and Altman and Brockman turned around, broke that trust, and commercialized it for billions. Tossing the case just because of a timeline, without even ruling on the merits of the actual betrayal, is a massive cop-out. Elon is absolutely right to take this to the Ninth Circuit. We see exactly what happened here, and the fight isn't over. We stand with you Elon.

We stand with you Elon. We stand with the truth.

Spot on. The era of the "thread-boi" is officially dead and buried in the source code. X basically weaponized Author Diversity Decay to break engagement pods by drastically diminishing returns on forced, repetitive interactions. Velocity is no longer a cheat code; it’s a liability if the quality isn't there.

@DrMelchisedecB Fascinating breakdown. The slop_score and Author Diversity Decay explain exactly why the old engagement pods and thread-boi templates completely fell off a cliff this month. Quality over velocity is finally being enforced by the codebase.

New X Algorithm: What X Didn’t Tell You (And What Actually Works in 2026) 🧵👇 While everyone else is guessing, I’ve spent the last 72 hours reading the entire X Algorithm code line by line since May 15, 2026. No guru opinions. No speculation. Just the actual code. Here’s exactly what stopped working after the May 15 Phoenix update, what’s winning right now, and what will still work months from now: What DIED (according to the code): 1. Spamming 4+ posts a day: The algorithm punishes you hard with *Author Diversity Decay*. Your 2nd, 3rd, and 4th posts get exponentially weaker reach in the same user’s feed. 2. Low-effort reply farming: Generic replies (“🔥”, “Great point”, emoji spam) get buried. Grok literally reads and ranks replies now. 3. Pure text-only posts: They work… but text + image or video (≥10 seconds) consistently outperforms them because of extra visual + dwell signals. 4. AI slop & recycled templates: There’s an explicit `slop_score` + quality classifier. Low-originality content gets flagged fast. 5. Cheap engagement bait: “What do you think?” or “Tag someone who needs this” increases scroll-past rate → active penalty via `not_dwelled` signal. What’s WINNING (and will keep winning): 1. Original, unique takes from small accounts: Original posts get full Grok evaluation + better out-of-network discovery. 2. Text + Media combos (especially video ≥10s with audio): Hybrid format maximizes multiple positive signals at once. 3. Contrarian opinions backed by real proof: Numbers, screenshots, personal results → higher dwell time, more replies & quotes. 4. Early, high-quality replies to big accounts (first 10-30 minutes): One of the strongest growth levers for small accounts right now. 5. Real conversation > likes: Quality replies + replying to your own comments in the first 30 minutes is pure ranking gold. My Day-One Test Results (Real Data): Reply 1 (contrarian + technical angle, +9 min after Elon’s post): → 214 impressions • 43 engagements • 39 detail expands • 2 profile visits. 👉[Link: x.com/DrMelchisedecB…] Reply 2 (thoughtful transparency take + direct question to Elon): → 45 impressions • 8 engagements • 0 profile visits 👉[Link: x.com/DrMelchisedecB…] No viral explosion. Just clean data. The real question isn’t whether 214 or 45 impressions is a failure. It’s whether you’re willing to treat it as data. Most people won’t. The ones who do are the ones who win. The Evergreen Playbook (Use This Today & Months From Now): - Cut volume to 1–2 high-quality posts per day max. - Always pair text with image, carousel, or short video. - Reply to every real comment in the first 30 minutes. - Lead with first-person stories + concrete proof. - One bold, specific opinion per post. - Focus on dwell time and real replies, not just likes. The Honest Question: Does this algorithm truly favor small accounts, or does it quietly give bigger accounts an unbeatable head start? It feels like a meritocracy with momentum, the rich get richer, but original, high-signal content from smaller accounts can still break through. I’m testing it publicly and aggressively for the next 30 days. If you want the full 72-hour technical breakdown (with every file reference, exact constants, codes, tables, and the hidden parts X didn’t publish), just say it in the comments below and I’ll post it in as reply to your comments. Save this post. The algorithm might change again in few months, but these principles won’t. I hope you've found this post helpful. Comment. Repost. Bookmark. Follow me @DrMelchisedecB for more... What’s your first move after reading this? Drop it below 👇

@Queenprissy19 Exactly. The data shows that the algorithm is finally shifting from rewarding "who you are" to "how long people stop and read." It’s a massive window of opportunity for small accounts before everyone else catches on.

@DrMelchisedecB A meritocracy with momentum" is the perfect way to describe X right now. Big accounts have the historical authority weight, but the Phoenix update showing love to out-of-network discovery for small accounts is a massive win if you actually have something unique to say.

@PianoEbute It looks like a lot, but you can boil it down to one simple shift: Quality > Quantity. Don't stress about mastering it all today. Just pick one thing, like focusing on making 1 high-quality post instead of 4 and start there. Which part surprised you the most?

@DrMelchisedecB Hmm!!! so much to study

I read through the published code. First, genuine respect to the X engineering team. Since Elon Musk took over, the technical quality of what's been built here is serious. The Phoenix transformer architecture, the Rust pipeline, the Grok integration, the two-tower retrieval system, this is world-class infrastructure. Publishing the code at all took confidence. Most platforms would never do it. The architecture is open. The weights aren't. I know likes add points and blocks subtract them, but not by how much. Grok scans every post for quality vs slop using classifier prompts that aren't published, so whatever biases are baked in, nobody can audit them. There's also a DO_NOT_AMPLIFY label X employees can attach to any account silently, no notification, no appeal anywhere in the code. The repo shows you the engine. It doesn't show you who's steering it or with what settings. A team this technically strong could publish the weight values and classifier prompts without exposing anything competitively sensitive. The skeleton is already out there. Releasing the muscle would make this the most transparent recommendation system ever shipped by a major platform. Any plans to publish the weights and classifier prompts, even partially, even versioned snapshots?

𝕏 now has over a billion downloads

Space-Based AI Infrastructure is the Only Way & Elon Musk Divergence on Humanity's Future🧵👇 Most people think Elon Musk talks about Mars because of ambition. That framing is outdated. His position and the physics behind it, is unsettling: Toby Ord warns civilization has a ~1-in-6 existential risk this century. • Ecological overshoot researchers warn Earth is already beyond sustainable carrying capacity. • AI leaders like Jensen Huang project a 1,000× compute explosion. • Thermodynamics says infinite growth on one planet ends in heat death or stagnation. • Planetary science says Earth itself has an expiration date. The uncomfortable truth is this: the timeline is compressing. What was once “eventually necessary” is becoming “within our lifetime necessary.” Elon Musk’s core claim remains uncompromising: Humanity is not energy-limited. Humanity is location-limited. And once that is understood, delay stops being neutral, it becomes a risk multiplier. Every year without expansion is a year closer to irreversible constraint. This analysis traces the first-principles argument across physics, entropy, AI scaling, solar evolution, existential risk, planetary limits and why space is no longer optional, but time-sensitive. Read the full analysis now👇 Space-Based AI Infrastructure is the Only Way & Elon Musk Divergence on Humanity's Future x.com/DrMelchisedecB… , why becoming multi-planetary is not science fiction, but a thermodynamic necessity. Earth is the cradle. The cradle is not the destination.

Yes, I need to make sure SpaceX stays focused on making life multiplanetary and extending consciousness to the stars, not pandering to someone’s bullshit quarterly earnings bonus! Obviously, IF SpaceX succeeds in this absurdly difficult goal, it will be worth many orders of magnitude more than the economy of Earth, but don’t expect entirely smooth sailing along the way.

SpaceX is intending to guarantee Elon Musk cannot be fired and offering a trillion-dollar pay deal linked to building a colony on Mars. ft.trib.al/byv6y9T

Missing features #NewXAndroidFeedback

@elonmusk Being able to drop a / and instantly pull in the right tool removes a lot of the friction that still exists in most chats. The skill-creator part is especially promising, it puts more agency in users’ hands. The direction is practical. Worth exploring. Exciting future ahead.

Grok now has skills

Impressive #1 on τ-Voice and the background reasoning with zero latency penalty is the real engineering story worth paying attention to. The Starlink deployment adds serious credibility: 70% autonomous resolution is great production signal. The true test now is consistency at scale on the messiest calls. Still early days yet great, the direction is crystal clear and the momentum is unstoppable. Solid step forward. Elon and his team are constantly breaking boundaries. Bravo!

Grok Voice is #1!

What was once a groundbreaking engineering miracle is now routine dominance, even on the challenging Pacific recoveries aboard OCISLY. Reusability isn’t just a feature anymore: it’s the foundation delivering lower costs, record cadence (268 consecutive successful booster landings and counting), and unmatched reliability for both national security payloads and commercial Starlink megaconstellations. SpaceX keeps making the “impossible” look effortless.

Falcon 9’s first stage lands on the Of Course I Still Love You droneship

Are we witnessing educational inclusion or educational decline? One side argues that rigid cutoffs in a deeply unequal country only exclude millions of students damaged by poor primary and secondary education systems. The other side worries that continuously lowering standards may normalize weak academic preparation instead of fixing the foundational collapse producing it. Nigeria’s education system seems degrading in relative global competitiveness. Lower admission thresholds without corresponding improvements in teaching quality, infrastructure, and skills training widen the gap with graduates from high-performing systems. This perpetuates a cycle of lower employability, brain drain, and reduced national innovation capacity. We must focus on foundational reforms (basic/secondary education), curriculum overhaul for 21st-century skills, increased funding, industry partnerships, and performance accountability. Without these, lowering bars helps short-term access but risks further diluting the value and global standing of Nigerian degrees. Nigeria has abundant human potential, the challenge remains systemic execution.

The Heads of Tertiary Institutions in Nigeria had unanimously agreed that the Minimum Admissible Scores for admissions into Universities should be 150, Colleges of Nursing, 150, and Polytechnics, 100.