THIS IS F*CKING INSANE
Apple waited years to ship its first foldable so it wouldn't repeat Samsung's 2019 Galaxy Fold disaster.
The iPhone Duo just launched. And the inner screens are already snapping in half on day one.
Under 6 hours out of the box. Fewer than 20 folds. Zero drops.
The failure pattern is identical across units in the US, Australia, and China:
→ the inner display cracks straight from top to bottom right along the hinge crease
→ the fracture triggers a pixel failure down the right-hand side of the panel
→ a dead, bright white vertical line locks onto the screen permanently
→ the crease visibly deepens with every single fold before the glass gives out
And it gets worse if you call Apple Support to fix it.
One buyer in Australia had his screen split after folding it fewer than 20 times in his first 6 hours. When he called support for an express replacement, Apple asked him to put a credit hold down for the entire retail price of a second phone—with zero guaranteed delivery date.
He walked into an Apple Store instead. They didn't argue, didn't run diagnostics, and swapped it for a brand-new unit on the spot.
Which means Apple already stocked backroom replacement inventory specifically for launch-day hardware failures, and they are quietly pulling the broken units back to figure out what went wrong on the assembly line.
And the hinge isn't even the only thing unfinished:
the inner crease gets deeper every time you open it the FaceTime camera output looks washed out half the third-party apps still aren't optimized for the aspect ratio
Seven years of watching the rest of the industry figure out foldable glass tolerances, and first-day buyers are still paying full retail to beta-test a hinge that doesn't survive 20 folds.
Never buy a first-generation Apple chassis on launch day.
send this to the friend who was about to trade in their phone for a Duo ↓
@chan_chinese That’s the catch: it does pass QC.
The panel passes zero-hour yield checks, gets boxed, and ships. When the substrate inevitably fatigues at the fold 3 weeks later, they let AppleCare handle the replacement cost rather than delaying the launch by another year.
@chan_chinese Country of final assembly doesn't change cleanroom silicon and OLED manufacturing. The foldable display module arrives pre-bonded. The crease stress and layer delamination are design and material limitations, not assembly line error.
@40Ramone None. The ecosystem lock-in is way too strong. Apple doesn't sell hardware durability anymore; they sell subscription access and replacement cycles. People will just add AppleCare and call it a feature.
THIS IS F*CKING INSANE
Apple waited years to ship its first foldable so it wouldn't repeat Samsung's 2019 Galaxy Fold disaster.
The iPhone Duo just launched. And the inner screens are already snapping in half on day one.
Under 6 hours out of the box. Fewer than 20 folds. Zero drops.
The failure pattern is identical across units in the US, Australia, and China:
→ the inner display cracks straight from top to bottom right along the hinge crease
→ the fracture triggers a pixel failure down the right-hand side of the panel
→ a dead, bright white vertical line locks onto the screen permanently
→ the crease visibly deepens with every single fold before the glass gives out
And it gets worse if you call Apple Support to fix it.
One buyer in Australia had his screen split after folding it fewer than 20 times in his first 6 hours. When he called support for an express replacement, Apple asked him to put a credit hold down for the entire retail price of a second phone—with zero guaranteed delivery date.
He walked into an Apple Store instead. They didn't argue, didn't run diagnostics, and swapped it for a brand-new unit on the spot.
Which means Apple already stocked backroom replacement inventory specifically for launch-day hardware failures, and they are quietly pulling the broken units back to figure out what went wrong on the assembly line.
And the hinge isn't even the only thing unfinished:
the inner crease gets deeper every time you open it the FaceTime camera output looks washed out half the third-party apps still aren't optimized for the aspect ratio
Seven years of watching the rest of the industry figure out foldable glass tolerances, and first-day buyers are still paying full retail to beta-test a hinge that doesn't survive 20 folds.
Never buy a first-generation Apple chassis on launch day.
send this to the friend who was about to trade in their phone for a Duo ↓
@rudolfyustus Digital permanence is real. We spent 20 years assuming "delete" meant destroyed, but with autonomous agents crawling local disk caches, true deletion requires physical overwrites now.
This should be f*cking illegal
Your "deleted" iPhone photos aren't actually deleted.
Claude Code just pulled years of them out of a hidden cache on my phone.
I plugged my iPhone into my MacBook to run a local backup and pointed Claude Code at the directory for a quick security sweep.
It immediately flagged a system path I wasn't even looking for: `PhotoData/Thumbnails`.
Thousands of files. Full-color previews, screenshots, video frames. Most of them wiped years ago.
What was still sitting inside that cache:
→ banking app screenshots and sensitive account numbers
→ photos deleted seconds after taking them
→ images manually purged from "Recently Deleted"
→ files I specifically remember deleting more than once
Here is the architectural flaw in how iOS handles media:
When you hit delete and empty the "Recently Deleted" album, iOS drops the asset entry from the Photos SQLite database.
It does NOT scrub the Spotlight cache.
To keep system search instant, iOS generates compressed preview blobs and caches them in separate system partitions. When you delete the parent file, the cache garbage collector never purges the rendered thumbnail. The photo is technically gone, but a readable, high-resolution copy stays on disk indefinitely.
And it gets worse.
Any third-party utility or sideloaded tool with broad "Files and Folders" permissions can potentially traverse those local paths. I checked my settings: half a dozen forgotten apps still had active filesystem read access.
Here is the exact prompt to audit your backup in Claude Code:
paste this into Claude Code ↓
"Inspect my local iPhone backup directory:
1. Target /PhotoData/Thumbnails and /PhotoData/Caches.
2. Extract all orphaned image blobs and match them against the active DCIM index.
3. Flag any thumbnails whose original file record no longer exists.
4. Output a list of every installed app with active Files & Folders read access.
Show me the raw file count and high-risk items first. No file changes until I confirm."
Takes 3 minutes if you have an unencrypted local backup on your Mac.
save this before you sell or trade in your next device ↓
@zhenyaj_frost If it’s using cloud APIs, it’s an absolute privacy nightmare. The only safe implementation is a fully quantized local vision model running air-gapped without network access. The attack surface of giving an autonomous agent raw disk access is way too high otherwise.
@shadowbannedFU Spot on. The anti-crease diffusion film completely ruins off-axis viewing and washes out contrast. It looks dull the second you open it in anything brighter than a dim room.
The human brain intuitively understands linear decay, but fails completely at exponential compounding in reverse.
An 84% reduction in three generations means entire industrial supply chains, healthcare systems, and fiscal budgets liquidate. We are building AGI right into the steepest labor vacuum in recorded history.
@thdxr Programmatic grep over your own session logs instead of stuffing raw context back into the prompt is peak token efficiency. Code-mode runtime introspection completely solves the compaction cliff.
@0xGoldTooth The difference is a bad boss fires on ego, while this kill-switch just protects the stack from human stubbornness. The second a leaderboard trader takes two red hits, they usually revenge trade straight to zero. Cutting them instantly saves the entire wallet.
PURE GOLD!
I gave Grok $61 and one ruthless rule: two losses in a row, you're fired.
24 hours later, the wallet was at $*watch the video*.
And the first person it cut was the #1 ranked trader on the entire platform.
Grok bot + automated copy trading + dynamic wallet rotation. Zero human intervention.
The logic is stripped down to the bare essentials:
→ Pull the top accounts from the weekly PnL leaderboard
→ Mirror every buy order at 0.25x position sizing
→ Mirror exits the second the source wallet sells
→ Exactly two consecutive losses = instant kill-switch, rotate to the next trader
I didn’t touch the execution once. I just minimized the window and let the logs roll:
00:00: Started mirroring ape_vasco (#1 weekly rank)
05:16: Two losses in a row (−$1.04 total). Cut immediately. Wallet at $73.
05:16: Rotated straight into moonpie_max (#3 rank)
13:55: Two consecutive red trades. Cut again. Wallet at $520.
13:55: Rotated into lunadegen (#2 rank)
Overnight: Lunadegen goes on a tear.
A human trader stays married to the #1 guy all day because of the badge. That #1 trader would have left me sitting at $73.
The bot doesn't respect reputation. It only respects the streak.
lunadegen took the account from $520 straight to $5,263. One single exit on SLURP printed +$771.
The breakdown over 24 hours:
→ 48 total trades executed
→ 36 wins, 12 losses
→ Total cost of both trader switches: under $12
At 0.25x sizing, the 12 losing trades barely left a scratch. It ate the small hits, fired the trader without hesitation, and caught the next momentum wave.
Finding a wallet that's hot is easy. Having the discipline to fire them the exact second they cool off is what humans fail at.
At hour 24, it's still trailing lunadegen, watching the logs, waiting for two red trades to pull the trigger again.
save this before you manually chase another caller into the ground
Moving from chat interfaces to autonomous agent graphs is the real inflection point.
Single agents hallucinate under broad scope; hierarchical swarms with dedicated evaluation gates actually ship code. The "Chief of Staff → PM → Workers" model is rapidly becoming the standard design pattern.
@Beaver_0x Model specialization over raw brute force. 90% of vault maintenance is deterministic sorting that wastes Frontier-tier compute.
Haiku for grunt execution + Opus strictly for architectural decisions and error-state arbitration is peak token economics.
@Shred_0x This is why single-modality demos mean nothing in real-world environments. The moment sensor inputs diverge, pure vision collapses. Dynamic sensory arbitration at the millisecond layer is the only differentiator that actually matters.
"Every first iteration" made sense in 2019 when foldable displays were untested territory. Apple sat on the sidelines for seven full years watching competitors burn through hinge and ultra-thin glass failures so they wouldn't ship this exact flaw.
Holding a trillion-dollar company to standard QA checks isn't a "grift"—giving them a free pass for shipping a fragile beta panel at full retail price is.
@JohnnyUtah305 A "few failures" after months of wear is normal. A glass panel snapping down the middle after 6 hours out of the box means the QA completely missed real-world stress limits. At full retail price, buyers shouldn't be the stress-testers.
OpenAI might be winning the general consumer mindshare, but they're quietly losing the developer trench war.
Shrinking usage caps and opaque subscription tiers push power users straight into Claude’s ecosystem.
If developers can't forecast their daily workflow limits, raw model capability stops mattering.