I Built a Transparent Katana

By Mike Shake · Diy · 25.8M views · 22:02

The teardown in brief

What's working

• The testing escalation (pineapple hardness 1 → brick hardness 10) is a textbook payoff ladder — every viewer can track exactly where they are in the sequence and the finish line is always visible
• The sword snapping after the pineapple test is the single best retention moment in the video — a genuine unexpected crisis that reframes every remaining test
• The distortion problem discovery at 9:00 is a well-handled mid-build roadblock that leads to a visible improvement and creates a natural comparison payoff between V1 and V2

What's costing attention

• Stakes are curiosity-only throughout — no consequence for failure is ever stated, which means the sword snap at 14:40 generates surprise but not crisis-level tension
• The polishing section (7:46-9:00) is the longest stretch of mechanical repetition in the build — seven identical passes with no intermediate reveals or checks compress into viewer fatigue
• The rebuild section (10:03-11:04) lacks a forward tease — 'let's make another sword' is honest but gives viewers no reason to believe version 2 will be worth the extra minutes

The first 30 seconds

have you ever seen an invisible sword well I think I have a few days ago I saw this picture right here that is an invisible sword and now I want one just like that a sword that's not all invisible but also destructible and sharp a sword that's basically like a normal one but that you can't see I'm going to build one no

Concept lands in 9 seconds, title promise is confirmed immediately, and the build plan is stated within 30 seconds — solid Tier 1 delivery, though the glass detour that follows softens the hold.

Where viewers drop

0:34 — Glass Detour Before Any Action (moderate)

You spend 54 seconds explaining why glass is bad before landing on polycarbonate. The viewer clicked for an invisible sword being built, not a materials science class. They're waiting for you to DO something.

Why it matters — You've already told us the premise at 0:09 — we're sold on the concept. Every second before the sledgehammer lands is time the viewer is wondering when the actual build starts.

8:00 — Seven-Pass Polishing Section (moderate)

You tell us you're repeating the same sanding pad process seven times — and the transcript confirms it with 'I'm going to repeat the process seven times and so I did.' This is a mechanical loop with no new stakes, no new information, and no intermediate reveals. The viewer is just waiting for it to be over.

Why it matters — The payoff (transparency) keeps getting deferred through this section. You've been promising the sword will look invisible since the hook — seven minutes into the build, viewers are still waiting and getting told to wait more.

9:24 — Rebuild Announcement Feels Like a Reset (moderate)

Three days of work gets partially abandoned. 'I already wrote the title of this video and I don't want to clickbait you' is an honest pivot but it arrives with zero forward momentum — you're essentially telling the viewer 'everything you just watched gets redone.' The rebuilt sword section (10:03-11:04) lacks urgency because you've already established what you're doing.

Why it matters — Without a clear tease of why version 2 is going to be better or a hint of what comes next, this section risks viewers deciding version 2 is just more grinding before the tests they actually want to see.

0:00 — No Consequences Ever Stated (moderate)

In 22 minutes, you never once say what happens if you fail. There's no bet, no promise, no consequence for a sword that can't cut anything. The viewer is curious but never scared — and curiosity alone doesn't hold retention as well as fear of a consequence.

Why it matters — When the sword snapped after the pineapple test at 14:40, it had the potential to be catastrophic tension. But because we had no stakes riding on that sword working, it's more of a surprise than a crisis. Consequences make every test matter more.

How the video is built

• 0:00 Concept & Material Validation
• 2:06 Build V1 — Shape, Bevel, Polish
• 9:01 Pivot — Distortion Problem & V2 Rebuild
• 13:14 Completion Reveal
• 14:00 Escalating Destructive Tests
• 21:49 Outro

What any creator can steal

• Add explicit consequences somewhere in the hook or before testing
• Cut or heavily compress the seven-pass polishing description
• Add a forward tease before the rebuild section
• Trim the glass explanation from the hook
• Add a forward tease at the end of the handle section before testing
• Film a 3-second 'final reveal tease' of your finished build at the very beginning of your shoot day — hold it up against the best background you have. This becomes your hook's payoff visual and your pre-testing trailer cut, and you'll have it in the can before you know how the rest goes.

More teardowns from Mike Shake

• How Dangerous is a Supersonic Nerf Dart?