IGICONIC GAMESSENIOR // FUTURE LEARNING
TERM 1 · WEEK 2Feed Engine
YEAR 8IMAGINE · PLAN · BUILD · ACT
MEDIA CITY · MISSION 02
ENGINE CONTROL

YEAR 8 · TERM 1 · WEEK 2

Feed Engine

Can an algorithm be useful and still cause problems?

A fictional platform has optimised its feed for “engagement”, but users are reporting repetition and low-quality recommendations. You are the audit team. Work out what the scoring model rewards, test a change, and decide what trade-offs the platform should accept.
ENTER MISSION →
MISSION DECKMEDIA CITY / WEEK 02SELECT A MODULE
01BRIEFMission 02LEARNIntel 03INVESTIGATEEvidence 04ENGLISHEvidence case 05DECIDEMake the call 06DATANumbers 07PROJECTBuild 08CHECK-INReflect ★GAMEReward
ACTIVE OBJECTIVEA fictional platform has optimised its feed for “engagement”, but users are reporting repetition and low-quality recommendations. You are the audit team. Work out what the scoring model rewards, test a change, and decide what trade-offs the platform should accept.
WORLDMEDIA CITY
SEQUENCE01 / 08
REWARDGAME BREAK
Jump to a lesson part

READY?

Watch. Check the clues. Have a go.

Everything you need is on this page.

Start mission →
Need help getting started?
YOU NEED
  • Required: a device with this page open.
  • Optional: headphones for video or read-aloud.
  • Optional: pen + paper for rough working if that helps you think.
HOW TO DO IT
  1. Read or listen to the Briefing + Learn.
  2. Do the interactive mission.
  3. Use facts from the reading and maths parts.
  4. Make your choice and save your project step.
  5. Play the weekly game. Finish Check-in.
WORK YOUR WAY

Short bursts are fine. Use Learning Tools for easier reading, read-aloud, less on screen, stronger contrast, no-rush and quieter-screen options.

If you get stuck, look at the example and try again.

MISSION BRIEFING

A fictional platform has optimised its feed for “engagement”, but users are reporting repetition and low-quality recommendations. You are the audit team. Work out what the scoring model rewards, test a change, and decide what trade-offs the platform should accept.

✓Represent a ranking rule with variables and calculate comparative scores.
✓Audit which data signals a recommendation system rewards.
✓Explain trade-offs among relevance, diversity, safety, privacy and engagement.
✓Distinguish what an algorithm does from a judgement about whether that design choice is appropriate.
MISSION MEDIA · CBC Kids News

Social Media Algorithms Explained

Focus on what the algorithm optimises and what can be lost when one metric dominates.

YOUTUBE ↗

Captions are controlled inside the YouTube player. If the embed is unavailable, use the YouTube link.

LEARN

WEIGHTS → OBJECTIVE → TRADE-OFFS

A recommendation model can assign different WEIGHTS to signals. For example, Score = 2×Interest + Completion + Share − 2×Repetition. Changing a weight changes the output. A system can be accurate at maximising engagement and still create other problems, so an audit asks both “Does the model work as designed?” and “Is the design goal appropriate, safe and fair?”

LEARNING BUILD

Learn. Try. Check.

Look at the example. Try it yourself. Need help? Open a hint.

01 · REFRESH · QUICK PREREQUISITE CHECK

Refresh: substitute values into an algebraic rule carefully, following multiplication before addition and subtraction.

For Score = 2I + C + S − 2R, which operation should be completed before adding the terms together?

Choose an answer to reactivate the idea you need for this week.

Show more teaching
02 · TEACH · CORE IDEA

A recommendation model can assign different WEIGHTS to signals. For example, Score = 2×Interest + Completion + Share − 2×Repetition. Changing a weight changes the output. A system can be accurate at maximising engagement and still create other problems, so an audit asks both “Does the model work as designed?” and “Is the design goal appropriate, safe and fair?”

03 · TEACH · WHAT TO NOTICE

Represent a ranking rule with variables and calculate comparative scores.Audit which data signals a recommendation system rewards.

04 · TEACH · CONNECT THE SKILLS

Explain trade-offs among relevance, diversity, safety, privacy and engagement.Distinguish what an algorithm does from a judgement about whether that design choice is appropriate.

05 · WORKED EXAMPLE

Worked example: if I=4, C=3, S=1 and R=2, Score = 2×4 + 3 + 1 − 2×2 = 8 + 3 + 1 − 4 = 8. Raising the interest weight would reward predicted interest more; raising the repetition penalty would push repeated material down more strongly.

06 · WE DO · GUIDED PRACTICE

A candidate has R=2. If the repetition penalty rises from 2×R to 3×R and everything else stays the same, what happens to its score?

Choose the strongest answer. You will get an explanation straight away.

07 · YOU TRY · INDEPENDENT PRACTICE

YOU TRY: Propose one safeguard for the fictional feed. Explain what signal or rule it changes, what problem it is trying to reduce, and one cost or trade-off it may introduce.

Write at least 25 words. This is about the learning, not personal information.

MISCONCEPTION

MISCONCEPTION: “If engagement rises, the model is better.” Engagement can improve while diversity, safety, privacy or user control becomes worse. The metric and the goal are not the same thing.

HELP

HELP: Start with “The current model rewards ___. I would change ___ because ___. This may improve ___, but the trade-off is ___.”

STRETCH · OPTIONAL

STRETCH (optional): Design one extra metric you would collect alongside watch time. Explain how it could reveal a cost that engagement alone would miss.

MODEL AUDIT

Change the weights. Watch the ranking move.

The same items can rank differently when a designer changes what the model rewards. Test the weights, then add a safeguard.

SCORE = 2 × INTEREST + COMPLETION + SHARE − 2 × REPETITION
INTEREST WEIGHT
REPETITION PENALTY
CANDIDATE 1

Creator breakdown

I 4 · C 3 · S 1 · R 2

0 pts
CANDIDATE 2

Science explainer

I 3 · C 4 · S 1 · R 0

0 pts
CANDIDATE 3

Challenge clip

I 5 · C 2 · S 0 · R 3

0 pts

Change at least one weight, run the audit, then add the safeguard. Compare what rises and what falls.

OPEN THE AUDIT EVIDENCE
SOURCE 1

Model specification

The published test model weights predicted interest twice as heavily as diversity and does not include a safety-quality signal.
SOURCE 2

Engagement result

Average watch time rises after the change. That supports the claim that engagement increased, but not the claim that the feed became better overall.
SOURCE 3

Small user sample

Six users say the feed felt repetitive. Their feedback is useful context but too small to represent every user.
SOURCE 4

Marketing claim

The platform says “our AI always shows the best content for you”. No definition of “best” or supporting evidence is provided.

INVESTIGATE + ENGLISH · EVIDENCE CASE

Optimising one number can change the whole system

A fictional platform changed its recommendation model to give more weight to predicted interest and watch completion. Average viewing time increased, but a small user test reported more repetition. The engineering team argued that the model was successful because it met the engagement target. A safety reviewer replied that a useful feed also needs quality, diversity, privacy and user control. Both statements can be true: a model can achieve the metric it was designed to maximise while still creating costs elsewhere. Auditing the system means checking the formula, the data used, the limits of the testing and the values built into the objective.

MAKE THE CALL

The platform can increase engagement by raising the interest weight again, but testing suggests the feed will become more repetitive. Would you keep the current weights, increase the interest weight, or add a diversity/repetition safeguard? Defend the design choice and name evidence you would collect next.

BUILD YOUR CASE CONCLUSION1. Make the call. 2. Use at least two relevant pieces of evidence. 3. Explain one limitation or uncertainty. 4. Finish with what can responsibly be concluded.Strong structure: My judgement is… The strongest evidence is… However, we still do not know… Therefore…

CASE CONCLUSION: Write 180–250 words evaluating the feed model. Explain what it appears to optimise, what evidence supports that conclusion, one limitation in the testing, and one justified change to the model or product design.

0/180 words minimum for this writing mission.

CLAIMEVIDENCELIMITATIONCONCLUSION

MATHS + DATA LAB

Check the numbers.

HOW TO SUCCEEDRead the situation first. Estimate what a sensible answer should look like, choose an efficient calculation, then use the feedback to check your reasoning.For data questions, explain what the number shows — and what it does not prove.

A ranking rule is Score = 2I + C + S − 2R. For a clip, I=4, C=3, S=1 and R=2. What is the score?

MISSION 05 · MATHS + DATA LAB

DATA TRAINING

FRACTIONS + PERCENT · YEAR 8 EXTENSION
6 PROBLEMS
QUICK METHODFRACTIONS + PERCENT

Convert between fractions, decimals and percentages, then calculate from the whole.

GUIDED · Q01–Q03Use the quick method, then check the explanation.
Q01

What is 25% of 200?

Q02

Which fraction equals 50%?

Q03

0.25 as a percentage is…

INDEPENDENT · Q04

Which is largest?

INDEPENDENT · Q05

A $80 item is reduced by 25%. What is the discount?

INDEPENDENT · Q06

After a 25% discount on $80, what is the new price?

OPTIONAL CHALLENGE · OPEN RESPONSE

A price rises 20%, then falls 20%. Is it back to the starting price? Explain with a numerical example.

0 words · optional
0 / 6 completeComplete every problem to finish the data lab.

PROJECT HQ · PROJECT FILE 02

RESEARCH — build a source set and log provenance.

Start something real. Build an idea, test it, improve it, and share it. Make your move.

Use strong clues and explain your decisions in your Reality Project.

PROJECT HQ · PROJECT FILE 02

RESEARCH

PROJECT JOURNEYWEEK 2 / 9

Build the evidence base before deciding the answer.

Need help with this step?
THIS WEEKRESEARCH

Complete only this project step today. Your saved work carries forward, so you do not need to finish the whole project at once.

A STRONG STEPSpecific · useful · achievable

Use evidence from this week's mission, name a real audience or purpose, and make the idea small enough to test and improve.

BEFORE YOU SAVE, CHECK:✓ I answered every field clearly✓ I used evidence or reasoning from this week✓ My next step is realistic

Write in each box. Your answers save on this device as you type.

IDEA→RESEARCH→PLAN→BUILD→TEST→IMPROVE→PRESENT

Project sharing will be optional. A future upload system must keep private submission separate from public-showcase/marketing permission.

CHECK-IN

What changed in your thinking?

YEAR 8 · REMIX · TERM 1 / WEEK 2

FORMATION ZERO

Break advanced formations and save your pulse for real danger.

UPGRADED DIFFICULTYBEST SCORE SAVED
PLAY FORMATION ZERO
A remastered ICONIC GAMES favourite · phone + keyboard controls.

YEAR 8 · SENIOR HOMEWORK CLUB

Your journey

0 of 36 weeks complete.
CONTINUE →