Skip to content
Home

Try it in 60 seconds

Most apps explain a concept. letsgye hands you the model to operate. Here are a handful, straight from real lessons, from Wi-Fi to problem solving, creativity, and seeing a problem many ways. Touch them.

Nothing you do here is sent anywhere. No sign-in, no install, no tracking. Every simulation runs in your browser, so go ahead and poke at it.

1 of 9

Why Wi-Fi drops through walls

Move the router, switch the band, and watch the signal fade through brick. Radio physics you can steer.

Pick a band, then move away from the router. Watch reach trade against speed.
Signal
77 %
Speed
501 Mbps

5 GHz is fastest up close, but watch it fade as walls get in the way.

2 of 9

How the internet survives a noisy line

Send a message and watch checksums catch the corrupted bits and ask for a resend. This is how files arrive intact.

Type a tiny message. Send it across a noisy network. Some bytes will get scrambled. How will the receiver know?
Sender
H72
E69
L76
L76
O79
Noisy network
hit Send to fly your packets across

Each byte has a 30% chance of getting scrambled in flight.

Receiver

Waiting for a packet…

Transmission log

Press Send to start a transmission.

What's happening in Python
message = "HELLO"
packet = message # no checksum yet
send(packet)
# at the receiver:
received = receive()
print(received) # no way to know if it's right

Try this: turn Checksum ON and crank noise to 60%. Most sends will fail, and now the receiver actually notices.

Then turn Retry ON. Watch the network try again until the message arrives clean. That tiny loop is the heart of TCP, the protocol your phone uses billions of times a day.

3 of 9

Ohm's law you can feel

Turn the dials and watch the voltage split. The same rule behind every dimmer, sensor, and charger.

Two-resistor divider: Vout = Vin x R2 / (R1 + R2). Switch R1 to an NTC thermistor to model a sensor.
+5VR110.0 kOhmA0R210.0 kOhmGND
Vout at A0
2.50 V
analogRead()
512
Current
0.25 mA
Vout = 5V x R2 / (R1 + R2)
= 5 x 10000 / (10000 + 10000)
= 2.500 V

Vout scales with R2 / (R1+R2). At R1=R2, Vout = Vin/2; R1 much greater than R2 approaches 0; R1 much less than R2 approaches Vin.

4 of 9

Why you forget, and what fixes it

Watch memory fade on the forgetting curve, then see how a well-timed review snaps it back. The science the whole app runs on.

How much you remember over 30 days. Turn on spaced reviews and watch the curve hold up.
100%50%0%day 01530
Retention at day 30
14%
At day 30
14%

With no review, retention crashes to about 14% by day 30. Turn on spaced reviews to see the curve hold up.

5 of 9

How a chart can lie

Slide the y-axis and watch the very same numbers look like a crisis, then like nothing. Spot the trick once and you see it everywhere.

Same numbers, two charts. Drag the axis off zero and watch one of them start to lie.
Honest (axis at 0)
0255075100Q1101Q2102Q3103Q4104
What they showed you (axis at 100)
100101102103104105Q1101Q2102Q3103Q4104
0 (honest)100 (chopped)
Real change Q1 to Q4
+3%
Q2 bar looks this much taller
2.0x
Q4 bar looks this much taller
4.0x

The right chart starts at 100, not 0, so the same +3% now looks huge: the Q2 bar looks 2.0x the Q1 bar, and Q4 looks 4.0x. The data did not move, the axis did. The eye reads bar heights, and chopping the axis stretches the gap.

6 of 9

Watch a problem get solved

Pick a strategy and watch it search the maze. Flood touches everything; Smart aims with a heuristic and reaches the goal through far fewer cells. Tap the grid to add walls and it re-plans on the spot.

Pick a strategy. Tap the grid to add walls.

The same idea routes your GPS, plans a chess engine, and moves a delivery fleet. A good heuristic is why a machine can weigh millions of options and still answer in a blink. The skill: search smart, not hard.

7 of 9

Grow an idea in every direction

Invent uses for an everyday object and watch them bloom. It counts how many, and how many different directions, then fuses two into something new. Creativity you can see, not a personality quiz.

Invent uses for a brick
0 ideas0 of 6 directions
Tap two dots to fuse them into a new idea.

Guilford's classic test of divergent thinking: fluency (how many) and flexibility (how many directions). It never scores an idea "good"; you go wide first, then choose. Fusing two is where novelty is born.

Train creative thinking for real
8 of 9

See one problem six different ways

Turn the lens on the same situation and watch it become a different problem each time, with the hidden assumption each view exposes. The skill behind good design, judgement, and invention.

0 / 6 lenses
Your prototype works in the lab but fails in the field.

Pick a lens. Watch the same problem become a different problem.

synthesis not yet

Each lens hides a different assumption. See at least 4 before you pick a side.

9 of 9

Kill the idea that would fail, cheaply

Every new idea rests on hidden assumptions. Spend a small budget to test the riskiest one first, watch the doubt collapse, and see what the same lesson would have cost after you built the wrong thing. Zero to one, de-risked.

You want to build a fees-payment app for your college.
Doubt still riding on the bet0% resolved
Students actually want an…The college will let a st…The logo and colours look…Students have a phone tha…
Students actually want an app for this
The whole idea assumes the current way (a queue at the office) is a real pain people would switch to fix.
The college will let a student app touch fees
Handling money means the administration and the bank must permit it. If they will not, nothing you build ships.
The logo and colours look polished
You could spend days perfecting the brand. It feels like progress and is safe to get right.
Students have a phone that can run it
Almost everyone on campus already owns a capable phone, so this is close to a settled fact.
Budget left 12 / 12 rupees

Real value-of-information: the drop in the doubt bar when you test an assumption is its weighted entropy (the information a clean test reveals), and a test is worth the fatal cost it saves you times the chance it fires, minus what the test costs. The cheapest, most load-bearing, most-likely-wrong assumption is almost always the one to test first.

Train zero-to-one judgment for real

That is one minute. There are hundreds more.

Every lesson is built like this: an accurate model you operate until the idea clicks, then spaced review so it sticks. Zero to expert, offline, and private.