Last lesson we buried the optimiser. Now we build the thing that replaces it. Herbert Simon’s alternative has a deliberately awkward name — satisficing, a portmanteau of satisfy and suffice — and a beautifully simple mechanism. You don’t hunt for the best. You decide, in advance, what “good enough” means, search until you find something that clears that bar, and then you stop and take it. This lesson makes that precise, works it on real numbers, and reveals a lovely fact: satisficing is not a vague attitude. It’s the exact same reservation-value stopping rule you met in the optimal-stopping course, wearing psychology’s clothes.
Before you read — take a guess
You're buying a used car. A satisficer's strategy is best described as which of these?
The two moving parts
Satisficing has exactly two components, and getting either wrong breaks it.
- The aspiration level. A concrete threshold that defines “good enough” — a bar an option must clear to be acceptable. For a flat: “two bedrooms, under £1,500, within 30 minutes of work.” For a hire: “clearly stronger than a solid mid-level engineer.” The aspiration is your standard, set (ideally) before you start.
- The stopping rule. Search options one at a time and accept the first one that clears the aspiration. The moment you find good enough, you’re done — you don’t keep looking to see if something even better is out there.
That’s the entire model. Notice what it doesn’t require: you never need the full option set, you never score everything, and you never compute a maximum. You need only be able to judge, one option at a time, does this clear my bar? — a question you can actually answer with bounded information.
Satisficing in one line
Set a bar for ‘good enough,’ search until something clears it, then stop. Two knobs — the aspiration level (how high the bar sits) and the stopping rule (take the first clearer). No full menu, no scoring everything, no maximum. Just: is this good enough? If yes, done.
A worked example: hiring a contractor
You need a contractor to remodel a kitchen. The optimiser’s fantasy — interview every contractor in the city, score each on price, quality, and timeline, and hire the global best — is both impossible (you can’t list them all) and absurdly costly (each meeting eats an afternoon). So you satisfice.
Set the aspiration. From two friends’ recent jobs and a bit of research, you decide “good enough” means: quoted under £22,000, at least 4 stars across 20+ reviews, and available to start within a month. That’s your bar.
Search and stop. You call contractors one at a time:
- Contractor 1: £25,000, 4.6 stars, available soon. Over budget — fails the bar. Skip.
- Contractor 2: £20,000, 3.9 stars, available soon. Rating too low — fails. Skip.
- Contractor 3: £21,000, 4.5 stars, starts in three weeks. Clears every part of the bar. Hire them. Stop.
You did not find the single best contractor in the city — there may well be a £19,500, 4.9-star gem three phone calls away. But you found one who is genuinely good enough after three calls instead of thirty, and you got your weekends back. Whether stopping was wise depends entirely on whether your bar was well set — which is the whole game, and where the next section goes.
Satisficing IS a reservation-value stopping rule
Here’s the connection that makes satisficing rigorous rather than folksy — and why this course sits downstream of optimal stopping. Recall the reservation value from that course: when you know roughly what quality to expect, you don’t sample to calibrate a bar — you set a threshold directly and accept the first option above it. That is exactly satisficing. The aspiration level is the reservation value. “Take the first option that clears the bar, then stop” is precisely the reservation-value stopping rule.
So everything you learned about reservation values transfers wholesale:
| Optimal-stopping language | Satisficing language |
|---|---|
| Reservation value (the acceptance threshold) | Aspiration level (the “good enough” bar) |
| Accept the first option above the reservation value | Take the first option that clears the aspiration |
| Threshold should fall as options run out | Lower your aspiration as time, money, and patience deplete |
| Search cost pushes you to stop earlier | Costly search → satisfice sooner, at a lower bar |
| Known distribution → set the value directly, no sampling | Experience tells you what “good enough” is worth demanding |
The one genuinely new idea satisficing adds is psychological: where does the aspiration level come from, and why do people cling to bars that are miscalibrated? Simon’s answer is that aspirations are adaptive — they rise when good options are plentiful and easy to find, and fall when the world turns stingy. Your bar for a dinner reservation is higher in a city full of restaurants than in a village with one pub. That adjustment is the descending (and ascending) threshold, driven by lived experience rather than a formula.
The descending threshold, made human
In optimal stopping, the reservation value falls mathematically as the horizon shrinks. In real life, your aspiration falls emotionally: the flat you’d have scorned in week one looks great in week six when the lease is up and you’ve seen twenty duds. That’s not you “giving up” — it’s your aspiration correctly tracking a shrinking runway and mounting search costs. Holding a rigid bar while the meter runs is how people optimise themselves into a worse outcome.
Setting the bar: the two failure modes
Because the aspiration level is a knob, it can be turned to the wrong place — and there are exactly two ways to blow it.
- Set it too LOW and you satisfice into junk: the first mediocre option clears your feeble bar and you stop, leaving huge value on the table. This is “good enough” as an excuse for not caring.
- Set it too HIGH and you never stop: nothing clears the bar, you search forever (paying mounting costs), and often get forced onto a poor last option anyway. This is the maximiser’s disease in satisficer’s clothing.
The lab makes both failure modes visible. The net-value curve humps: low aspiration is bad (junk), high aspiration is bad (endless costly search), and there’s a sweet spot in between. Drag the aspiration to the far left — quality craters. Drag it to the far right — watch the “looks used” balloon and net value collapse. The peak is a well-set bar.
Satisficing lab
A well-set aspiration captures most of the value
Options arrive one at a time with a hidden quality (1–100). A satisficer sets an aspiration level and grabs the first option that clears it; a maximiser inspects everything to find the true best. Drag the aspiration, then raise the cost of each look and watch which strategy wins.
- Accepted quality
- 89
- Looks used
- 4.6
- Regret
- 8
- Net value
- 79
| Quality | Looks | Net value | |
|---|---|---|---|
| Satisficer | 89 | 4.6 | 79 |
| Maximiser | 97 | 25 | 47 |
With 25 options and an aspiration of 78, the satisficer accepts quality 89 after just 4.6 looks for a net value of 79. The best aspiration here is about 81. The maximiser finds the true best but pays for every look — its net value is only 47.
Watch one search
Notice how the peak moves as you change the cost per look. When looking is cheap, you can afford a higher bar (keep searching for something great). When each look is expensive, the best aspiration drops — good enough, sooner. This is the reservation value responding to search cost, exactly as optimal stopping predicts. A well-set aspiration isn’t a fixed number you memorise; it’s a bar you tune to how plentiful the options are and how costly the search.
In the lab, cranking up the 'cost per look' moves the best aspiration level to a LOWER number. Why?
Why this is rational, not lazy
The crucial reframe: satisficing is optimal subject to constraints. Given that you can’t list every option, can’t score an uncertain future, and pay for every look, “set a good bar and take the first clearer” is not a compromise on rationality — it is the rational procedure. The optimiser who insists on the global best is the irrational one here, because they’re pursuing a target that’s impossible to hit and expensive to chase.
But — and this matters — “rational subject to constraints” still demands that the bar be well set. Satisficing is not a licence to lowball your standards and call it wisdom. The skill the whole course is teaching is precisely this: pick the right aspiration for the situation, tune it to the options and the costs, and then have the discipline to actually stop when it’s cleared.
Match each satisficing concept to what it actually means.
Pick a term, then click its definition.
The one thing to remember
Satisficing = an aspiration level plus a stopping rule: set a bar for good enough, take the first option that clears it, and stop. It’s the exact same reservation-value machinery as optimal stopping — the aspiration is the reservation value, and it should fall as search gets costly or time runs short. It’s rational subject to real constraints, not lazy — but only if the bar is well set. Too low, you settle for junk; too high, you never stop. Next we turn to the psychology: why the people who refuse to satisfice — the maximisers — often end up worse off.