Four corrections, each one an error you make because you're competent. Then you retake the baseline and find out what actually moved.
Five weeks of this have been about access — starting at all, having something to say, sustaining it, getting in and out, and pointing your attention outward. Quality was deliberately postponed, because quality at zero volume is worth nothing. You should now have volume. This week is the first one about doing it better rather than doing it more.
All four corrections share a structure worth naming up front: each is a mistake that people make in proportion to how capable and considerate they are. None of them are laziness errors. They're all overshoots of something reasonable, which is why they're invisible from the inside and why they've survived twenty years of you being fine at work.
You ask a question. They answer. You pivot to your own answer, which is what you wanted to say all along.
The numbers are unkind. Boomeraskers enjoy the exchange more than their partner does (5.46 vs 4.91). Recipients read it as insincere — worse than plain self-disclosure, because the question created an expectation of interest that the pivot then withdrew. And 83% of people believe wrapping a disclosure in a question makes a better impression than just disclosing. It doesn't: 71% vs 56% second-date acceptance, in favour of just saying the thing. Brooks & Yeomans 2025
If you want to share something, just share it. That's the whole correction. The question-wrapper isn't buying you the permission you think it is — and it's costing you the sincerity you didn't know you were spending.
This one is yours specifically, and it connects straight back to Lesson 2. Your original complaint was that you have "no reason to speak" — that a contribution needs a licence issued from outside. Boomerasking is what that belief looks like once you're already in a conversation. You manufacture the licence by asking first, so that your disclosure arrives as a response rather than an offering. It's the same move as waiting for a pretext, relocated inward, and it's the last place that belief is hiding.
Someone tells you something good that happened to them. There are four ways to respond, and only one of them does anything.
| Active-constructive | Enthusiastic, elaborating, asking for detail. The only one that predicts relationship quality. |
| Passive-constructive | Warm but brief. "Oh, nice." Patterns with the destructive responses. |
| Active-destructive | Pointing out the downside. "Won't that mean a longer commute?" |
| Passive-destructive | Ignoring it, or changing the subject. |
Look at row two, because that's where you'll be. "Oh, nice" is polite. It is not rude, not dismissive, not cold. And it groups with the destructive responses in the outcome data. The failure mode here is politeness — restraint, not indifference — which makes this the easiest high-leverage change in the entire curriculum. You don't have to stop doing something harmful. You have to stop under-doing something harmless.
Short one, because you already have it. Week 2's novelty penalty applies to stories just as it does to openers: given the choice, speakers pick the novel story 66.7% of the time, and listeners react worse to it, d=1.10. Cooney, Gilbert & Wilson 2017
The reason it belongs in a quality week rather than staying back in Week 2: as your volume goes up, so does the temptation to reach for your best material. Don't. The familiar story lets the listener fill in the gaps themselves, and that filling-in is the part they actually enjoy. Your interesting story asks them to do homework first.
People overestimate how awkward a deep conversation will be by d=2.23 — roughly four times the awkwardness that's actually experienced. That is an enormous effect, one of the largest in this whole curriculum, and it runs in the direction that costs you conversations you'd have valued. Kardas, Kumar & Epley 2022
Disclosure turns negative in public settings with strangers. Depth works once a conversation is established — it is not an opener, and deployed as one it produces exactly the reaction you've been afraid of all along. The sequence is: get in (Weeks 1–4), establish the thing (Week 3's follow-ups), then go past where it feels comfortable. Order matters here more than in any other item on this list.
Catch your boomerasks. Count them, don't suppress them — same logic as the filler track: you can't drop a habit you can't yet detect. Log each one and what you actually wanted to say. Trending to zero across the week.
Upgrade every piece of good news you hear. Whenever someone reports something that went well, respond active-constructively: enthusiasm plus one request for detail. That's it. Target four a day, which is fewer than you'll be offered.
One deliberate depth question a day, in a conversation that is already running — never as an opener, never cold in public. Something you'd normally hold back as too much.
Three days of active-constructive responses. Tap as you go.
| Boomerasks caught | Per day, trending to zero. A rising count on day two is normal and means detection improved, not that you got worse. |
| Active-constructive | Responses given. Target 4/day. |
| Depth attempts | One a day, plus a one-line note on how awkward it actually was versus how awkward you predicted. You're re-running Lesson 2's prediction gap on a different variable. |
Jelly Jam, Lisbon's English-language improv jam — beginner-friendly, with randomly assigned partners. The random assignment is the useful property: it removes the selection step entirely, so everyone you talk to is someone you didn't choose, which is the closest thing available to a controlled test of everything above.
At the end of this week, do two things.
Retake the baseline. The same six-item measure you took before Week 1, on the Charisma: Baseline card. This is why you took it early — a self-report score six weeks in is nearly meaningless on its own and quite informative as a difference.
Comparing against Week 1: has my initiation latency actually dropped — or have I just gotten better at conversations I was already having?
Be honest about that second possibility, because it's the likelier failure and it feels like success from the inside. Weeks 3, 5 and 6 all improve conversations that are already underway. Only Weeks 1, 2 and 4 attack the thing you originally came here with, which was never conversation — it was initiation. If the latency number from Week 1 hasn't moved, the programme has produced a real improvement in the wrong variable, and what you'd want is another cycle weighted heavily toward Weeks 1 and 2, not a new curriculum.
From memory: name the four corrections, say what they have in common, and give the final checkpoint question.
Stop boomerasking; respond active-constructively to good news; tell the familiar story; go deeper than feels comfortable, but only once a conversation is established. What they share: each is an overshoot of something reasonable — a mistake made in proportion to competence and politeness, not despite them. The checkpoint: has my initiation latency actually dropped, or have I just improved conversations I was already having?
Brooks, A. W., & Yeomans, M. (2025). "Boomerasking: Answering Your Own Questions." Free PDF from HBS
Read it for the gap between what boomeraskers think they're doing and what recipients receive — it's the cleanest example in this curriculum of a move that feels considerate from the inside and reads as self-interested from the outside. If you only act on one item from this week, though, make it the good-news response: it's the smallest change with the largest documented payoff, and Gable et al. 2004 is also free.