For six lessons you used encodings other people invented. Now you invent one — and find out what it costs.
Build your own scheme for turning letters into bits. Use it to encode a message. Hand it to a partner who has to get your exact message back out.
If they can't recover your message, your encoding didn't work — and that is useful information, not a failure. Step 4 asks you to diagnose it, and diagnosing it correctly is worth the same points as getting it right the first time.
Do the arithmetic before you assign a single code. This is the whole design decision, and it is the same counting skill from 1.1: every bit you add doubles the patterns available.
You must answer three things: how many symbols your message needs, the smallest number of bits that works, and — this is the part people skip — why the next size down does not work. "5 bits" with no pattern count is half credit.
A bit pattern for every symbol. Same length for all of them. No duplicates. This table is your encoding — hold onto that phrase, you'll need it in the writing.
8 to 15 characters, something a classmate would find worth decoding. Group the bits by symbol so you can check your own work, then count: characters × bits each = total bits.
Hand over your table and your bits. Say nothing else. Copy your partner's bits down before you start decoding, so you have a record of what you were actually given.
If it fails, say specifically where: was the problem in their table, their encoding, or your decoding? "It didn't work" earns nothing. "Their table has B and R both as 00001" earns full marks.
ASCII spends 8 bits per character. Work out your message both ways and find the difference.
Then the question the whole project is built around: ASCII is bigger. So why does every computer on Earth use ASCII instead of your scheme? Think hard about what your partner needed from you in Step 4.
Work Steps 1 and 3 out on paper first. Then use this to check yourself — it will tell you whether your reasoning holds, not what to write.
One paragraph, 5–8 sentences. Budget real time for it. You must use all three of these words correctly:
Your paragraph must answer all four:
Half the project is the paragraph. Plan your two days accordingly.
Big Idea 2 questions about encoding almost always come down to one skill: counting patterns. "How many bits are needed to represent N things" and "how many things can n bits represent" are the same question asked in both directions, and they are guaranteed points if you're fluent. Step 1 is that exact question with your own message attached to it.
The last two rubric rows train something else. The exam's written responses reward describing a tradeoff precisely, and punish claiming more than your evidence supports. "Smaller than 8-bit ASCII for this message, at the cost of losing lowercase" scores. "More efficient" does not.
Give your most common letters shorter codes. It saves bits, and it introduces a trap.
If A=0, B=1, and C=01, then the bits 01
could mean A-then-B, or it could mean C, and nobody can tell which. A variable-length
code only works if no code is the beginning of another code.
Try E=0, T=10, A=110, O=111 and decode 0110100 — it has
exactly one possible reading. Then explain in two sentences why the fixed-width scheme you
built on page 1 never has this problem.