
How to compare options with Super Intelligence is not about asking AI which choice is “best”. It is about making the comparison structure visible enough that you can see what each option gives you, what it costs, what assumptions support it and which conditions would make the answer change.
The strongest comparison does not begin with the options. It begins with the decision.
What are you actually trying to achieve? Which requirements are non-negotiable? Which preferences matter but can be traded? Which evidence is current? Which uncertainties could flip the choice? Which option exists only because you forgot to consider waiting, repairing, simplifying or doing nothing?
In the eduKateSG life series, Super Intelligence, or SI, is our editorial name for practical AI assistance. This guide follows How to Use Super Intelligence to Challenge Your Assumptions and prepares the ground for decision matrices and second-order thinking.
The aim is human-owned comparison: SI organises evidence and trade-offs; you decide what matters and what consequence you are willing to accept.
The First Comparison Error: Starting with the Options
Suppose you ask SI to compare Laptop A and Laptop B. The system can produce a detailed specification table. But if your real need is light travel, long battery life and dependable support, half the table may be irrelevant. If your current laptop already meets those needs, both options may be unnecessary.
The comparison should therefore begin with the target outcome and constraints before products, courses, schools, jobs, routes or services are named.
Ask: “What problem am I trying to solve, and what would make an option unacceptable even if it scores well elsewhere?”
This prevents the available options from defining the question.
The Comparison Spine
Use this sequence: Decision → Requirements → Options → Evidence → Trade-offs → Uncertainty → Sensitivity → Choice → Review.
Decision
State the actual choice in one sentence.
Requirements
Separate must-haves from preferences.
Options
Include genuinely different routes, not cosmetic variations.
Evidence
Attach current facts and sources to material criteria.
Trade-offs
Show what each option gains and sacrifices.
Uncertainty
Label what is estimated, assumed or unverified.
Sensitivity
Test which criteria or assumptions can flip the choice.
Choice
Make the human decision at the level supported by evidence.
Review
Define what future evidence should cause reconsideration.
Must-Haves, Preferences and Bonuses
Must-haves
If an option fails a must-have, it is normally removed before scoring. Examples include required software compatibility, budget ceiling, schedule compatibility, accessibility or legal eligibility.
Preferences
These can be traded against each other: lighter weight versus larger screen, salary versus commute, live lessons versus flexible timing.
Bonuses
These are nice features that should not outweigh the core decision: free accessories, decorative dashboards or extra modules you may never use.
Ask SI to label your criteria rather than treating every feature as equally important.
Disqualifiers Come Before Scores
Scoring an option that cannot meet a hard requirement creates false precision.
A course that clashes with a non-negotiable work schedule should not receive points for excellent content and then appear competitive overall. A product that cannot run required software should not remain because it is cheaper.
Build the elimination gate first. Score only options that remain feasible.
The Missing Option Problem
People often compare A versus B when the best decision space contains C, D or no action.
Always ask SI to generate at least one structurally different option: delay, repair, rent, borrow, reduce scope, change process, stay with the current state, run a pilot or combine parts of two options.
The hidden option frequently changes the decision more than a more detailed comparison between the original choices.
Criteria Should Come from the Decision, Not the Product Page
Product pages, course brochures and job descriptions present criteria favourable to themselves. Your comparison criteria should come from your actual use case.
A laptop seller emphasises processor, display and design. You may care more about battery, repairability, warranty, weight and software compatibility.
A course provider emphasises content hours and certification. You may care more about prerequisite fit, practice quality, schedule and whether the skill is demonstrable.
A job advertisement emphasises responsibilities. You may care more about decision authority, learning, commute, manager quality and stability.
Ask SI to derive criteria from your goal and constraints before reading marketing attributes.
Evidence Columns Matter
A comparison table should distinguish facts from estimates and judgements.
- criterion;
- option value;
- source;
- date checked;
- confidence;
- assumption;
- human judgement or note.
This stops a subjective score from looking like an externally verified fact.
Current Information Needs Current Sources
Price, availability, software support, course schedules, eligibility, opening hours and policies can change. The comparison should record when time-sensitive facts were checked.
A generated comparison is not current merely because it was generated today. SI may be reasoning from old context or a source that has changed.
For consequential decisions, verify material facts with the issuing provider or another suitable primary source.
Total Cost, Not Sticker Price
Cost is usually multidimensional. A device has purchase price, accessories, software, repair, downtime and replacement risk. A course has tuition, materials, travel, time and displaced opportunities. A job has salary but also commute, benefits, work hours and flexibility.
Ask SI to build a total-cost model appropriate to the decision.
Time is a cost
Do not convert every hour into money automatically, but record time when it constrains the choice. A cheaper option that consumes two additional evenings each week may be more expensive in the dimension you actually value.
Maintenance is a cost
An option requiring constant configuration, cleaning, updates or supervision should not be compared only on acquisition cost.
Switching is a cost
Migration, learning, setup and disruption can matter when replacing a current system.
Opportunity Cost
Every chosen option displaces something else.
Taking a second course may displace a work project. Buying the premium product may displace savings. Choosing the far-away job may displace family time. Scheduling another activity may displace recovery.
Ask SI: “What does each option prevent me from doing with the same time, money or attention?”
Opportunity cost is often invisible because the unchosen alternative never appears on a receipt.
Reversibility Changes the Comparison
A reversible option can be attractive even when it is not perfect because you can learn and change course.
A one-month trial, pilot project, refundable booking or temporary schedule can generate evidence. An irreversible purchase, contract, resignation or major commitment deserves stronger confidence.
Add a reversibility criterion: how difficult is it to undo, and what does reversal cost?
This often favours experiments over premature full commitments.
Option Value Under Different Scenarios
An option can be best in one future and weak in another. Instead of forcing one forecast, compare options across scenarios.
Example: a powerful laptop is attractive if heavy local computation becomes necessary. A lighter model is better if work remains browser-based. A third option—delay purchase—may dominate while workload is uncertain.
Ask SI to build a scenario table and identify which option remains acceptable across the widest plausible range.
Robustness can matter more than maximum performance in one forecast.
Sensitivity Analysis
Sensitivity asks which criterion or weight can flip the preferred option.
Suppose Job A wins when commute matters moderately, but Job B wins if commute is weighted heavily. That tells you the choice is not objectively A; it depends on how important commute is to you.
Ask SI: “Vary one criterion at a time. At what point does the preferred option change?”
This exposes fragile decisions and directs attention toward the values or facts that deserve more thought.
Dominance and Pairwise Comparison
Sometimes one option is no worse on every important criterion and better on at least one. That option dominates the other under the current model. Remove dominated options before complex scoring.
When many options feel overwhelming, compare two at a time on the most important criteria. Pairwise comparison can expose why one option is preferred without forcing a large matrix immediately.
If pairwise comparisons become cyclic—A beats B, B beats C, C beats A—the criteria or trade-offs need to be made explicit.
Regret and the Cost of Being Wrong
Two options with similar expected value may have very different downside. Ask: “If this choice is wrong, which error is easier to live with?”
A reversible modest purchase may create low regret. A high-cost course with little transferability may create larger regret. A job decision can be compared by how recoverable each route is.
Regret is subjective and should not replace evidence. It is useful when downside asymmetry matters.
A Comparison Table That Does Not Lie
Criterion | Importance | Option A | Option B | Option C | Source | Confidence | Notes.
If a value is unknown, write unknown. If a score is subjective, label it subjective. If a criterion is a hard requirement, mark pass/fail instead of allowing a high score elsewhere to compensate.
The table should expose uncertainty rather than force every cell into a number.
Do Not Average Away a Fatal Weakness
A weighted average can make a disqualified option appear attractive. An unsafe product, impossible schedule, missing eligibility condition or unacceptable privacy risk should not be rescued by points for convenience.
Use gates before scores. This principle becomes even more important in the next article on decision matrices.
The Comparison Narrative
After the table, require a short narrative explaining why the decision changes under different priorities.
Example: “Option A is preferable when initial cost and portability dominate. Option B becomes preferable when future local performance and repairability receive greater weight. Delay is preferable while workload requirements remain uncertain.”
The narrative prevents the table from becoming an unexplained verdict.
Start with the Decision, Not the Shopping List
Option comparison goes wrong when the alternatives arrive before the decision criteria. A person sees three courses, four laptops or two job offers and immediately begins comparing visible features. The comparison becomes anchored to what the options advertise rather than what the person actually needs.
The stronger sequence is: define the decision → define the requirements → define disqualifiers → gather evidence → compare trade-offs → test sensitivity → choose.
This sequence matters because options can manipulate the criteria simply by being available. A premium laptop highlights screen brightness and processor speed. A course highlights certification and hours of content. A job offer highlights title and salary. Those features may or may not determine your real outcome.
Ask SI first: “Before comparing these options, help me define the decision and the criteria that would still matter even if these particular options disappeared.”
That prompt protects the comparison from becoming a feature contest.
The Comparison Architecture
A robust comparison has nine layers.
1. Decision question
What are you actually choosing? “Which laptop is best?” is vague. “Which laptop best supports four years of travel-heavy writing, browser research, video calls and light local data work within my budget?” is a decision.
2. Must-haves
Requirements an option must satisfy to remain eligible. Examples: software compatibility, a maximum commute, a required qualification, a confirmed date, a price ceiling.
3. Disqualifiers
Conditions that eliminate an option regardless of other strengths. A course may be excellent but unusable if the timetable conflicts with a non-negotiable commitment.
4. Preferences
Criteria that improve fit but are negotiable: lower weight, nicer interface, more flexible schedule, preferred location.
5. Evidence
What source supports each comparison point? Current official specifications, contract terms, direct observation, verified reviews, personal tests.
6. Uncertainty
Which criteria are estimated, subjective or not yet verified?
7. Trade-offs
What do you gain and lose with each option?
8. Sensitivity
Which assumption or criterion would change the choice if its importance moved?
9. Review trigger
What new evidence would justify reconsidering the decision later?
Must-Haves, Preferences, Bonuses and Disqualifiers
A comparison becomes clearer when criteria are not all placed on one flat list.
Must-haves
If the option fails a must-have, it is not eligible. Do not allow a high score elsewhere to compensate unless you intentionally relax the requirement.
Preferences
Preferences can trade against one another. A slightly heavier laptop might be acceptable for much longer battery life. A longer commute might be acceptable for significantly better role fit.
Bonuses
Nice-to-have features that should not drive the decision. Bonuses often receive disproportionate attention because they are easy to market.
Disqualifiers
Conditions that should eliminate the option immediately: incompatible software, unaffordable total cost, timetable conflict, missing required accreditation, unacceptable risk.
Ask SI to classify every criterion into one of these categories before scoring anything.
This reduces the common error where an attractive bonus outweighs a failed requirement.
Evidence Quality in Comparisons
An option comparison is only as good as the evidence filling it.
For current products, use current specifications and prices. For courses, use current provider information and verify accreditation or eligibility where relevant. For jobs, distinguish the job description from assumptions about culture. For travel, distinguish confirmed bookings from suggested itineraries.
Use an evidence label beside important criteria:
- verified current source;
- direct experience;
- reliable secondary source;
- estimated;
- subjective preference;
- unverified claim.
A table filled with unverified claims is not rigorous because it has columns.
Ask SI to flag where one option has stronger evidence than another. Missing evidence should remain missing rather than being guessed.
Worked Example 1: Comparing Two Courses
Decision: choose one course for the next six months without undermining the current workload.
Must-haves: relevant to target skill, fits available evenings, total cost below budget, provides enough practical work to demonstrate learning.
Preferences: live feedback, recognised provider, shorter commute or online access.
Disqualifier: requires three evenings a week when only two are sustainably available.
Course A costs less and has a recognised certificate. Course B costs more but includes a project reviewed by practitioners.
A weak comparison might ask which course has better reviews.
A stronger comparison asks which course produces the capability the learner wants to demonstrate. If employer evidence suggests project work matters more than the certificate, Course B may become more attractive. If the learner cannot realistically complete the project work, that advantage collapses.
Sensitivity question: “If certificate recognition is less important than I currently assume, does the preferred option change?”
Opportunity-cost question: “What would I have to stop doing to complete each course well?”
Reversibility question: “Can I defer enrolment without losing something material?”
The final decision is therefore not “A versus B”. It is “Which route produces the needed skill under the actual constraints?”
Worked Example 2: Comparing Two Job Offers
Decision: choose between Offer A and Offer B for the next stage of a career.
Visible features include salary and title, but the relevant criteria may include autonomy, learning, manager quality, commute, stability, promotion pathway, working hours and role scope.
Start with personal criteria before reading the offers again. This reduces anchoring to salary or title.
Now label evidence. Salary is confirmed. Commute can be estimated. Manager quality may be partly inferred from interviews but remains uncertain. Promotion pathway may be described but not guaranteed.
Ask SI to separate confirmed terms from inferred culture.
Then run scenario analysis:
Scenario 1: manager quality is average in both roles.
Scenario 2: Offer B’s manager is excellent.
Scenario 3: commute becomes more burdensome than expected.
If the preferred option changes under small uncertainty, the decision is sensitive. Gather more evidence before committing if practical.
If the same option remains preferable across plausible scenarios, the decision is more robust.
Worked Example 3: Comparing a Major Purchase
Decision: choose a laptop for four years of study and work.
Must-haves: required software, budget, acceptable battery, minimum storage, warranty in the country of use.
Preferences: weight, screen quality, port selection, keyboard feel.
Bonus: premium finish.
Disqualifier: software incompatibility.
Now calculate total cost rather than sticker price. Include necessary adapters, warranty differences, likely repair cost and whether extra performance will actually be used.
Ask SI to create a “no-purchase” option: keep current laptop and upgrade only if a specific workload appears.
This is important because comparison systems often assume one of the listed products must win.
The no-purchase option protects against buying to resolve uncertainty that could be tested first.
Worked Example 4: Comparing School Support Options
Decision: choose the next support approach for a student whose Mathematics performance has declined.
Options: more group tuition hours, one-to-one diagnostic support, targeted self-practice with feedback, or maintain current support while changing the practice method.
Criteria: demonstrated bottleneck, fit with student’s schedule, independence, feedback quality, cost, ability to retest.
If diagnostic evidence shows one narrow algebra prerequisite is unstable, general extra hours may be less targeted than one-to-one repair or a structured practice sequence.
If the student cannot maintain independent practice at home, the more targeted option may still fail operationally.
The comparison therefore includes both instructional quality and execution feasibility.
SI can help structure the options, but actual student work should determine the diagnosis.
Opportunity Cost: Every Option Includes What You Give Up
An option is not just its benefits and price. It also consumes resources that could serve another option.
A course consumes evenings that could be used for the current project.
A cheaper apartment farther away may consume travel time.
A high-paying job may consume flexibility.
An automation project may consume maintenance attention that could improve the underlying process.
Ask SI: “For each option, what valuable alternative use of time, money or attention becomes unavailable?”
Opportunity cost is often the hidden criterion that changes the choice.
Reversibility and Option Value
Options differ in how easy they are to reverse.
A one-month trial subscription is more reversible than a one-year contract. A short internal project is more reversible than quitting a job. Borrowing or renting equipment is more reversible than purchasing.
When two options are otherwise close, reversibility can be valuable because it preserves future choice while uncertainty remains.
Ask SI to identify the most reversible path that generates evidence.
This is especially useful when the user is trying to make a large decision with information that will improve over time.
Uncertainty Should Be Visible in the Comparison
Do not fill uncertain cells with confident estimates merely to make the table complete.
Use labels such as confirmed, likely, estimated, unknown or subjective.
Then ask which unknowns could change the choice.
If an unknown is low impact, proceed. If it is load-bearing, investigate.
Example: uncertain commute
Two jobs appear similar, but one commute is uncertain because the office location may change. If commute is a major value criterion, this uncertainty has high decision impact. Confirm it or model both locations.
Example: uncertain course workload
A course advertises six hours a week but past students report ten. If available time is tight, workload uncertainty may be more important than small differences in curriculum.
Uncertainty belongs in the decision, not outside it.
Sensitivity Analysis
Sensitivity analysis asks whether the choice changes when assumptions or weights change.
You do not need advanced mathematics.
Change one important factor at a time.
If price matters twice as much, does the choice change?
If commute is 20 minutes longer, does the job ranking change?
If the certificate has little employer value, does the course ranking change?
If the device lasts three years instead of five, does the total-cost conclusion change?
An option that remains preferable across reasonable changes is robust.
An option that wins only under one narrow set of assumptions needs more evidence or a more explicit value judgement.
Dominated Options
Sometimes an option is worse than another on every material criterion.
If Option B costs more, performs worse on the required task, has weaker warranty and offers no unique benefit, it is dominated by Option A.
Remove dominated options before complex scoring.
This keeps the comparison focused and prevents irrelevant choice overload.
Ask SI: “Are any options clearly dominated once must-haves and verified criteria are applied?”
The Pareto Frontier
When no option dominates another, the best choices may lie on a frontier of trade-offs.
One laptop is cheaper and heavier. Another is lighter and more expensive. Neither is universally better.
One job pays more but offers less flexibility. Another pays less but reduces commute and offers stronger learning.
The frontier tells you where the real decision lives: among options that trade different strengths.
SI can identify the frontier, but the user supplies the values that select along it.
The No-Regret Option
A no-regret option performs reasonably well across several plausible futures.
It may not be optimal in any single scenario, but it avoids large downside if assumptions are wrong.
For uncertain situations, ask SI to identify which option is robust rather than which has the highest expected upside.
This is useful for career experiments, technology purchases and planning under changing schedules.
Comparing Qualitative Criteria Without Pretending They Are Precise Numbers
Some criteria resist honest numerical scoring.
Manager quality, learning fit, family convenience, creative freedom and sense of trust may matter greatly but cannot be measured precisely before the decision.
Do not assign 8.3/10 merely to make the matrix look scientific.
Use qualitative labels with evidence: strong, adequate, weak, unknown.
Then explain why.
If a numerical decision matrix is useful, keep subjective scores visibly subjective and test whether small score changes alter the choice. Article 15 in this series develops that method in depth.
The Evidence-Asymmetry Problem
One option may have more information available simply because it is more familiar.
A current job has months of evidence. A new job has an interview and a description. The current school has direct experience; a new school has public information. Your current device has known weaknesses; a new product has marketing claims.
The familiar option can look worse because its problems are visible, while the unfamiliar option looks cleaner because its problems are unknown.
Ask SI to mark where evidence quantity differs and avoid treating missing negatives as evidence of absence.
Unknown is not the same as good.
Comparison Traps
Anchoring
The first price, score or option frames the comparison. Repair by defining criteria before examining options in detail.
Feature counting
An option wins because it has more features, even though most do not matter. Repair with must-haves and bonuses.
Single-metric domination
Salary, price, ranking or speed overwhelms other relevant criteria. Repair by stating the actual outcome.
False precision
Subjective judgements are converted into precise scores. Repair by labelling uncertainty and running sensitivity analysis.
Option omission
The real best route is delay, repair, rent, borrow, simplify or do nothing. Always include the outside option.
Evidence mismatch
One option is supported by current official information and another by anecdotes, yet both cells appear equal. Add evidence quality.
Winner-first reasoning
The user prefers one option and builds criteria to justify it. Repair by defining criteria independently and asking SI to test the preferred option hardest.
Comparison without action
The user keeps adding options and criteria after the decision is clear enough. Add a stopping rule.
A Complete Comparison Protocol
Step 1: write the decision in one sentence.
Step 2: define the outcome the choice should produce.
Step 3: list must-haves and disqualifiers.
Step 4: list preferences separately.
Step 5: include the no-action or delay option.
Step 6: gather current evidence for each material criterion.
Step 7: mark unknowns and evidence quality.
Step 8: remove dominated options.
Step 9: write the major trade-off for each remaining option.
Step 10: calculate opportunity cost and reversibility.
Step 11: run sensitivity analysis on the two most important uncertain criteria.
Step 12: decide or identify the one missing fact that blocks the decision.
Step 13: record the choice and review trigger.
This protocol can be done conversationally or converted into a decision matrix when the comparison is large enough to justify one.
A Copyable Comparison Prompt
“Help me compare these options without recommending immediately. First restate the decision. Separate must-haves, disqualifiers, preferences and bonuses. Include a no-action or delay option if relevant. For each material criterion, label the evidence as verified, estimated, subjective or unknown. Remove dominated options. Then show the key trade-off for each remaining option, opportunity cost, reversibility and which assumption would most change the choice. Run a sensitivity check on that assumption before recommending anything.”
A Full Comparison Case: Choosing Between Renting and Buying Equipment
Decision: whether to buy a specialised device needed for occasional projects.
Must-have: supports required software, available when needed, total annual cost acceptable.
Options: buy new, buy used, rent per project, use a shared facility, delay until project volume rises.
Evidence: current purchase prices, rental rates, expected project frequency, maintenance cost, resale uncertainty.
Assumption: project frequency will increase from two to eight uses per year.
Sensitivity: if frequency stays at two, renting dominates. If it rises above six, buying may become cheaper. If shared access is reliably available, the threshold moves again.
Reversibility: renting preserves flexibility. Buying used may preserve resale value. New purchase creates the largest commitment.
Decision: rent for the next two projects and revisit after actual demand is observed.
The comparison produces an experiment rather than forcing a purchase under uncertainty.
A Full Comparison Case: Choosing a Revision Method
Decision: how a student should revise vocabulary over the next month.
Options: reread lists, flashcards with retrieval, sentence writing, mixed application, SI-generated quiz.
Must-have: requires active retrieval and transfer into writing because the goal is usable vocabulary, not recognition alone.
Rereading fails the must-have and is removed as the main method.
Flashcards support retrieval. Sentence writing supports production. Mixed application supports transfer. SI quizzes can provide variation but require answer checking.
The correct comparison is not “which method is best?” It may be a sequence: retrieval → sentence use → mixed transfer.
This illustrates another comparison lesson: sometimes the options are complements rather than substitutes.
Substitutes, Complements and Sequences
Before comparing, ask whether the options actually compete.
A tutor and an AI practice system may be complements.
A cheaper laptop and a cloud-computing subscription may be a bundle.
A diagnostic assessment and a study plan may be sequential.
Comparing complements as though only one can survive creates a false choice.
Ask SI to classify options as substitute, complement or sequence. This can change the architecture of the decision completely.
When Comparison Should Stop
Stop when:
- only one option satisfies the must-haves;
- remaining differences are below the importance threshold;
- one option dominates the others;
- the choice is robust across reasonable sensitivity changes;
- the missing information would cost more to obtain than the decision warrants;
- a reversible trial will produce better evidence than further analysis; or
- the decision deadline has arrived and the evidence is sufficient for the consequence.
Comparison is a means to action. Endless comparison can become a way to postpone responsibility.
The Comparison Review
After the decision has had time to produce evidence, review the criteria.
Which criterion mattered more than expected?
Which feature was irrelevant?
Which uncertainty resolved differently?
Which source was most trustworthy?
Which assumption should become a reusable check next time?
Did the chosen option perform well because the comparison was good, or because circumstances happened to favour it?
This review makes future comparisons faster and better calibrated.
A Reusable Comparison Sheet
Decision: [what is being chosen]
Outcome: [what the choice should achieve]
Must-haves: [requirements]
Disqualifiers: [elimination rules]
Preferences: [tradeable criteria]
Options: [including no-action/delay]
Evidence: [source + freshness]
Unknowns: [what may change the choice]
Trade-off: [gain/loss per option]
Opportunity cost: [what each option displaces]
Reversibility: [how easy to undo]
Sensitivity: [which change flips the choice]
Decision: [selected route]
Review trigger: [new evidence or date]
What Progress Looks Like
- you define criteria before becoming attached to an option;
- must-haves are separated from bonuses;
- disqualifiers remove unsuitable options early;
- unknown information stays visible;
- evidence quality is not hidden by neat tables;
- opportunity cost appears in the comparison;
- reversibility is valued when uncertainty is high;
- the no-action option is considered;
- you test how sensitive the choice is to assumptions;
- you recognise when options are complements rather than substitutes; and
- you stop comparing when the decision is clear enough to act.
Worked Comparison 5: Travel Route A, B or C
Route A has the lowest fare but a long overnight connection. Route B costs more and arrives in daylight. Route C uses another airport with cheaper ground transport but less schedule frequency.
Compare total journey cost, door-to-door time, baggage, connection risk, sleep disruption, refundability, airport transfer and usefulness of arrival time.
A low fare can become a high total cost when hotel, transfer and lost day are included. If the traveller is arriving for an important morning event, robustness and sleep may dominate fare. If travel is flexible and budget is the hard constraint, Route A may remain attractive.
The comparison belongs to the whole journey, not the airline ticket.
Worked Comparison 6: Home Service Providers
Three providers quote different prices for a repair. Normalise scope before comparing. One quote includes parts and follow-up. Another excludes parts. A third includes a longer warranty.
Criteria may include qualification where relevant, scope, total price, availability, warranty, response time and evidence of reliability.
SI can create a standard comparison sheet and identify missing terms to ask each provider. The cheapest headline quote may disappear once exclusions are made visible.
Worked Comparison 7: Build, Buy or Subscribe
A small business needs a software capability. Options: build internally, buy a perpetual product, subscribe to a service, or use a manual workflow temporarily.
Compare setup, recurring cost, maintenance, customisation, integration, lock-in, security requirements, staff skill and time to value.
A build option can look cheaper when internal staff time is excluded. A subscription can look expensive when maintenance burden is ignored. A manual process can be rational while volume remains low.
The decision should scale with actual need rather than imagined future size.
Worked Comparison 8: Exercise or Wellness Routines
A person compares gym membership, home exercise, classes and a simple walking-plus-strength routine.
SI can organise schedule, cost, travel, enjoyment, equipment and adherence constraints. It should not claim medical suitability for an individual without appropriate professional input where needed.
The strongest programme is not automatically the one with the most features. It is the one that fits the person’s circumstances well enough to be used safely and consistently.
This example shows how comparison criteria can include behaviour and practical fit rather than only theoretical performance.
Hybrid Options
Do not assume you must choose one option exclusively.
A learner can combine one structured course with a real project. A family can use public transport most days and occasional ride-hailing when time-critical. A business can keep a manual process for exceptions while automating the routine path.
Ask SI to generate hybrid options only after the main trade-offs are visible. Otherwise hybrids can become complicated bundles that hide rather than solve the decision.
A good hybrid addresses a specific weakness of one option using a limited part of another.
Sequential Options
Some choices can be staged. Repair first, then replace if the problem returns. Pilot first, then scale. Take a short course, then decide on certification. Rent first, then buy.
Sequential options convert uncertainty into evidence over time.
Add a transition rule: what result moves you from Stage 1 to Stage 2? This prevents temporary solutions from becoming permanent by inertia.
Portfolio Options
Some decisions are not one-shot choices. You can allocate resources across several options.
A learner can allocate most time to the current syllabus while preserving a small exploration block. A household can keep most savings in a stable buffer while allocating a smaller amount to a higher-risk goal. A professional can deepen one core skill while testing a second area through a small project.
SI can help model allocations, but the user should state the risk and priority boundaries first. A portfolio should reduce concentration risk without becoming so fragmented that nothing receives enough commitment.
The Do-Nothing Baseline
Every comparison should ask what happens if you keep the current arrangement.
Doing nothing can carry cost: ongoing inefficiency, missed opportunity, increasing repair expense or continued frustration. It can also preserve optionality and avoid premature commitment.
Write the baseline explicitly: current monthly cost, current time burden, current failure rate, current benefit and expected trend if nothing changes.
This prevents a new option from appearing attractive merely because the existing state was never measured.
The Delay Decision
Delay is not neutral. It has an information value and a waiting cost.
Ask two questions: what useful information will arrive if I wait, and what do I lose by waiting?
If a new product release, confirmed timetable or medical appointment will materially change the decision soon, waiting may be rational. If the current problem is causing compounding harm, delay can be expensive.
SI can map the waiting trade-off without pretending future information is guaranteed to arrive.
The Pilot Decision
A pilot is especially useful when the option is uncertain but reversible.
Define duration, scope, success evidence, failure evidence and what decision follows. A one-month workflow test, a trial class, a short rental, a shadowing project or a prototype can turn assumptions into data.
Do not allow the pilot to become permanent by default. Set an end date and a review decision.
The Comparison Audit
Decision: What exactly am I choosing?
Outcome: What problem must the choice solve?
Hard constraints: What cannot be violated?
Option set: Have I included delay, no action, repair, pilot or hybrid routes where relevant?
Criteria: Do they come from my use case?
Overlap: Am I double-counting one preference?
Evidence: Which values are sourced and current?
Unknowns: Which gaps can change the result?
Total cost: Money, time, maintenance and switching?
Opportunity cost: What does each option displace?
Reversibility: How expensive is being wrong?
Scenarios: Which future states matter?
Sensitivity: Which weight or assumption flips the choice?
Decision: What am I choosing, and why?
Review trigger: What future evidence should reopen it?
This audit is enough for most meaningful personal comparisons without pretending the process is mathematically objective.
The Comparison Exit Test
- Have infeasible options been removed?
- Are the remaining options genuinely different?
- Are must-haves separate from preferences?
- Are current facts verified where consequence requires it?
- Are subjective judgements labelled?
- Have total cost and opportunity cost been considered?
- Have delay, no-action or pilot options been considered?
- Do I know which criterion or assumption could flip the result?
- Is the preferred option robust enough for the uncertainty?
- Can I explain the choice without simply repeating a model score?
If yes, the comparison is probably ready for decision or for the more formal matrix in the next article.
Common Comparison Failure Modes
Too many criteria
The table becomes noise. Keep criteria that can change the decision.
Criteria overlap
Battery life, mobility and travel suitability may double-count the same underlying preference.
Scoring before elimination
An infeasible option remains because its average score looks high.
Marketing criteria
The seller’s feature list defines the comparison instead of the user’s need.
Stale evidence
Current prices or schedules are assumed from old information.
False objectivity
Subjective preferences are disguised as precise numbers.
Missing do-nothing option
The user compares purchases without considering repair or delay.
Ignoring opportunity cost
Time or money displaced by the choice remains invisible.
Single-scenario thinking
One uncertain future determines the whole choice.
No sensitivity test
The user never learns which weight or assumption flips the result.
Recommendation dependence
The user asks SI to choose before making personal values explicit.
A 10-Minute Comparison Prompt
“Help me compare these options. First restate the decision and identify missing options, including delay or no action. Separate hard requirements from preferences. Remove options that fail a hard requirement. For the remaining options, create criteria based on my use case, not marketing features. Label facts, estimates and subjective judgements. Include total cost, reversibility, opportunity cost and important uncertainty. Show what assumption or priority would flip the preferred option. Do not make the final decision for me.”
A 21-Day Comparison Practice
Days 1–3: practise separating must-haves from preferences.
Days 4–6: add hidden options such as repair, delay and pilot.
Days 7–9: check evidence provenance and freshness.
Days 10–12: compare total cost and opportunity cost.
Days 13–15: build scenario tables and test robustness.
Days 16–18: perform sensitivity checks on criterion weights.
Days 19–20: write a comparison narrative instead of relying on a score.
Day 21: save one reusable comparison audit in your personal SI system.
The objective is not to compare everything exhaustively. It is to recognise which structure makes a particular decision clearer.
Comparisons Need a Reference Point
A comparison without a reference point can make every option look attractive. Decide what “good enough” means before evaluating alternatives. The reference point may be the current situation, a minimum acceptable standard, the best existing option or the cost of doing nothing.
For example, a new productivity app should not be compared only with competing apps. Compare it with the current workflow after simple cleanup. If the existing system becomes adequate once duplicated lists are removed, the new app must clear a higher bar.
In education, compare a proposed tuition option with the student’s current baseline and the specific bottleneck, not with an abstract idea of “more support”. If the existing tutor already targets the real gap, switching providers may add transition cost without improving the mechanism.
Ask SI: “What is the correct reference point for this comparison, and how would the decision change if I compared every option against the current baseline rather than against each other?”
Threshold Criteria: When a Preference Becomes a Requirement
Some criteria are not simply more-is-better. They have thresholds.
Battery life may be irrelevant above eight hours if that already covers a full workday. Commute time may be acceptable below forty-five minutes but sharply unattractive above an hour. Tuition cost may be manageable up to a family budget ceiling and unacceptable beyond it.
Ask SI to identify threshold criteria before scoring. Once an option clears the threshold, extra performance may have little practical value.
This prevents premium features from receiving excessive weight simply because they are numerically larger.
Comparing Bundles Instead of Single Items
Real choices are often bundles.
A laptop decision includes accessories, warranty, software and possible cloud costs. A school support decision includes tuition time, travel, homework load and family scheduling. A job decision includes salary, commute, benefits, manager, learning and future options.
If only the headline item is compared, hidden bundle costs distort the result.
Ask SI to decompose each option into its complete bundle: direct cost, recurring cost, setup cost, maintenance, required companion tools, time commitment and exit cost.
The correct comparison may reveal that the apparently cheaper option is more expensive once the bundle is included.
Comparing Sequences
Some options should not be chosen once and for all. The best strategy may be a sequence.
Career example: test management through one cross-team project → review fit → then decide whether to pursue a formal management role.
Learning example: diagnostic support → targeted repair → independent practice → decide whether more tuition hours are required.
Technology example: use current device → rent specialised hardware for one project → buy only if repeated demand appears.
Ask SI: “Can I convert this either-or choice into a sequence that preserves options and produces better evidence?”
Sequential comparison is especially useful when uncertainty is high and the irreversible choice can be delayed.
The Regret Test
Regret is not a perfect metric, but it can reveal asymmetry.
Ask two questions: “If I choose A and B turns out better, what do I regret?” and “If I choose B and A turns out better, what do I regret?”
The answers may reveal that one mistake is far more costly than the other.
For example, delaying a course by three months may have little downside if enrolment is frequent. Taking it now and burning out may harm the current course and family time. The asymmetry can make delay more attractive even if the expected benefits are similar.
Do not let regret replace evidence. Use it to expose asymmetric downside that simple scores may hide.
The Worst-Case and Best-Case Bounds
When inputs are uncertain, compare reasonable bounds.
For a purchase, calculate total cost under optimistic and pessimistic lifespan assumptions.
For a course, compare workload under advertised hours and a higher realistic estimate.
For a commute, model normal and peak conditions.
For a project, compare completion under expected and delayed approval.
If an option remains acceptable across the range, it is robust. If it becomes unacceptable under a plausible downside scenario, the uncertainty deserves more attention.
Ask SI to use ranges when the inputs are estimates instead of fabricating one precise number.
The Minimax Regret Lens
When probabilities are unclear but downside matters, one approach is to minimise the largest plausible regret rather than maximise a guessed expected value.
Suppose you can buy a specialised device now or rent until demand is clearer. Buying could save money if usage becomes frequent, but create a large unused asset if demand remains low. Renting may cost more under high usage but keeps the downside bounded.
A minimax-regret framing asks which choice avoids the largest painful mistake across plausible futures.
This is not automatically the right decision rule. It is useful when the user values downside protection and reliable probabilities are unavailable.
Ask SI to present several decision rules—expected value, robust choice, minimax regret—and show whether they point to different options.
The Option Value of Waiting
Waiting can have value when information will improve.
If a product refresh is imminent, a course timetable will be released next month or a job’s internal scope will become clearer after a project, delay may produce information without large cost.
But waiting also has cost: lost time, missed enrolment, continued inconvenience or foregone opportunity.
Ask SI to compare the value of information gained by waiting with the cost of delay.
This turns “wait or act?” into a structured comparison instead of treating delay as indecision.
Comparing Under Budget Constraints
A budget is not merely another criterion. It can reshape the option set.
If an option exceeds the hard budget, either remove it or explicitly decide that the budget itself can change. Do not allow a high overall score to quietly override a financial boundary.
For soft budgets, show the marginal benefit gained per additional dollar.
If Option B costs 40% more but improves a low-priority criterion by 5%, the premium may be hard to justify. If it eliminates a major risk or satisfies a must-have, the premium may be rational.
Ask SI to explain the incremental benefit between adjacent options rather than only scoring each one independently.
Comparing Under Time Constraints
Time has the same structure as money.
A course requiring eight hours weekly may be objectively strong but unsuitable for a person with four sustainable hours. A job with a higher salary may consume an extra ten hours of weekly travel. A cheaper purchase may require extensive maintenance.
Convert time into the actual weekly or monthly burden. Include travel, setup, recovery and administrative overhead.
The best option on paper can fail because its time bundle does not fit the user’s life.
The Pairwise Comparison Method
When there are many options, comparing everything at once becomes cognitively difficult.
Use pairwise comparison.
First remove disqualified and dominated options. Then compare the remaining options two at a time on the most important criteria.
Ask: “Between A and B, which trade-off matters? Between the winner and C, which new trade-off appears?”
Pairwise comparison is useful when qualitative criteria dominate and a giant matrix creates false precision.
However, pairwise results can become inconsistent—A beats B, B beats C, but C beats A. When this happens, inspect whether different criteria are driving each pair and make the value weights explicit.
The Ranking Stability Test
A ranking is stable when small reasonable changes do not reorder the top options.
Ask SI to vary one uncertain criterion at a time. If the top two keep switching, the decision is fragile. If one option remains ahead across most plausible settings, it is robust.
Fragility tells you where to research next. Do not spend time improving evidence for criteria that never affect the ranking.
This is one of the strongest ways SI can reduce research overload: identify which uncertainty actually has decision value.
The Decisive-Information Test
Ask: “What single missing fact would most likely change the choice?”
For a job, it may be actual working hours rather than another salary comparison.
For a course, it may be whether assessed project work receives real feedback.
For a purchase, it may be whether required software runs reliably.
For school support, it may be the student’s actual diagnostic error pattern.
Research that fact first.
A good comparison system directs effort toward decisive information rather than accumulating detail everywhere.
The Uncertainty Budget
You do not need to eliminate every unknown.
Allocate attention according to decision impact.
High-impact unknowns deserve verification. Medium-impact unknowns may deserve ranges or scenario analysis. Low-impact unknowns can remain unresolved.
Ask SI to sort uncertainties by expected decision impact. This makes research proportional to consequence.
The uncertainty budget is especially useful when a comparison is approaching a deadline and not every question can be answered.
A Full Comparison Case: Choosing a Tuition Format
Decision: choose the most appropriate support format for a Secondary student with a diagnosed algebra bottleneck and limited weekly time.
Options: large group tuition, small-group tuition, one-to-one tutoring, structured self-practice with periodic tutor review.
Must-haves: targeted algebra repair, regular feedback, independent practice, sustainable weekly schedule.
Preferences: short travel, predictable cost, peer interaction.
Disqualifier: format spends most time following a general class sequence without addressing the diagnosed prerequisite.
Evidence needed: class size, lesson structure, how diagnostics are used, actual timetable, travel time, cost and how homework feedback works.
Now compare mechanism, not marketing.
Large-group tuition may offer structure and lower cost but less individual diagnostic time. One-to-one may offer focused repair but higher cost and fewer peer cues. Small-group may balance both if the group is genuinely small and work is differentiated. Self-practice may be efficient when the learner is already independent but weak when follow-through is the bottleneck.
Sensitivity: if independence improves after two weeks of targeted repair, the need for high-intensity support may fall. Reversibility: a short diagnostic block before a long-term commitment preserves options.
The comparison therefore may produce a sequence: short targeted one-to-one repair → small-group or self-practice once the bottleneck stabilises.
The correct output is not a universal winner. It is a fit model linked to the student’s demonstrated needs.
A Full Comparison Case: Choosing a Travel Route
Decision: choose between a direct but expensive flight and a cheaper one-stop itinerary for a family.
Criteria: total cost, total travel time, connection risk, baggage handling, arrival time, children’s tolerance, cancellation protection.
Must-have: arrival before a fixed event. Disqualifier: connection window below the family’s minimum acceptable buffer.
The cheaper flight saves S$600 but adds five hours and a connection.
Opportunity cost includes fatigue on arrival and a more fragile schedule.
Sensitivity: if the direct-fare difference drops to S$300, preference may flip. If the connection includes a long safe layover and the next day is free, the cheaper option may become acceptable.
The comparison should therefore include the schedule around the flight, not only fare and duration.
This illustrates the system principle: options belong inside the life they affect.
The Comparison Handoff into a Decision Matrix
A decision matrix is useful after the criteria and evidence are already clean.
Do not begin with a spreadsheet and invent criteria to fill it.
First complete the comparison architecture: must-haves, options, evidence, uncertainty and trade-offs. Then use a matrix when multiple criteria need consistent weighting or when the reasoning must be shared with others.
The next article, How to Build a Decision Matrix with Super Intelligence, develops scoring, weighting, sensitivity and matrix failure modes in detail.
A matrix should compress a good comparison, not substitute for one.
The Final Comparison Preflight
Before deciding, ask:
- Did I define the decision before examining features?
- Does every remaining option satisfy the must-haves?
- Did I include a no-action, delay, repair or trial option?
- Are current facts sourced and uncertainty labelled?
- What is the main trade-off for each option?
- What does each option displace?
- Which choice is most reversible?
- Which assumption could flip the decision?
- Is the ranking stable under reasonable changes?
- What missing fact has the highest decision value?
- Are any options complements rather than substitutes?
- Is the decision clear enough that more comparison now has low value?
If the answers are visible, the comparison has probably done its job.
Frequently Asked Questions
How can SI help compare options?
It can structure criteria, normalise information, identify missing options, expose trade-offs, test sensitivity and distinguish facts from assumptions.
Should I ask SI which option is best?
For meaningful personal decisions, it is usually better to define criteria and values yourself and use SI to organise the comparison.
How many options should I compare?
Enough to cover genuinely different routes. Include delay, repair or no action when relevant.
What are must-haves?
Requirements an option must satisfy to remain feasible. Options failing them should usually be removed before scoring.
Should I use scores?
Scores can help when criteria are explicit, but they should not hide subjective judgement or allow fatal weaknesses to be averaged away.
What is sensitivity analysis?
It tests whether changing a criterion weight, assumption or scenario changes the preferred option.
Why include reversibility?
Reversible options can create evidence and reduce the cost of being wrong.
How do I compare uncertain information?
Label confidence, use scenarios and verify material current facts. Do not force unknowns into precise scores.
What is opportunity cost?
The value of what you give up by choosing an option, including time, money, attention or another project.
What comes next?
The next guide shows how to build a decision matrix without turning subjective judgement into fake mathematical certainty.
Helpful Reading
- How to Leverage Your Life with Super Intelligence
- How to Use Super Intelligence to Challenge Your Assumptions
- How to Think More Clearly with Super Intelligence
- How to Use Super Intelligence to Break Down Complicated Problems
Compare the Decision, Not Just the Options
A good comparison does not produce a winner by force. It reveals which options are infeasible, which criteria matter, which facts are uncertain, which trade-offs you are accepting and which assumption could change the answer.
Super Intelligence should make the comparison more inspectable—not make the human decision disappear.
Continue with How to Build a Decision Matrix with Super Intelligence when the comparison needs explicit weights, scores and sensitivity analysis.
Use This Guide Inside the Super Intelligence Learning Curriculum
This page is also the canonical Article 34 owner in the How to Learn Super Intelligence Quickly series. Use it after critical-thinking practice to turn assumptions and evidence into explicit comparison criteria, and before later decision-making articles that address uncertainty, trade-offs and human choice in more depth.
The protected method is the same across personal and learning use: define the decision, separate hard requirements from preferences, compare options on the same criteria, preserve unknowns, test sensitivity and keep the final choice with the responsible human.
Critical Thinking → Compare Options → Better Ideas
In the SI learning curriculum, use Critical Thinking first to clarify claims, evidence and assumptions. Use this page next to compare the resulting options on common criteria. Then continue to Generate Better Ideas With Super Intelligence when the current option set is too narrow.
