Lesson Report:
# Title
**Forecasting with Baselines, Signals, and Presentation Planning**
This session began with a rapid review of the forecasting framework introduced last week, especially the shift from asking whether an event will happen to specifying the conditions under which it could happen. The class then moved into a more concrete method for weighting signals and costs, using student case studies on U.S.–Iran nuclear escalation, U.S. control of Greenland, and a possible Chinese blockade of Taiwan. The final part of the lesson focused on next week’s 5-minute forecasting presentations, including structure, professionalism, and submission expectations.
—
# Attendance
– **Students explicitly mentioned absent:** **0**
– **Absent student names mentioned:** None in the transcript
—
# Topics Covered
## 1) Opening review: forecasting as probability, not certainty
– The instructor opened by apologizing for the delay and immediately moved into a **brief review of last week’s work on forecasting**.
– He restated the core principle that **analysts cannot say with certainty that something will or will not happen**; instead, all claims must be framed **probabilistically**.
– The review emphasized that the course is moving from a general understanding of probability toward a **more concrete forecasting method**.
– The instructor referenced concepts introduced previously, including:
– **costs**
– **signals**
– **cheap talk**
– **baselines**
## 2) Reconstructing last Thursday’s exercise: from outcomes to conditions
– The instructor asked the class to recall the main task from the previous lesson: rather than asking only **“Will this happen?”**, analysts should ask **“Under what conditions would it happen?”**
– One **uncertain student** raised an important conceptual distinction between:
– **“under what conditions would it happen”**
– **“what are the conditions for it to happen”**
– The instructor used this question to clarify that the goal is to build a **list of conditions that would need to align** for an event to become more likely.
– He noted that this is **not absolute prediction**:
– analysts cannot know every condition in advance,
– and not every listed condition must be fulfilled with total certainty,
– but **the more relevant conditions that are met, the more the likelihood increases**.
– The example used throughout was intentionally extreme: **the United States dropping a nuclear bomb on Iran**.
– Students were asked to think through **what would need to happen first** for such an outcome to become plausible.
## 3) Review of the probability line and the meaning of the first dot
– The instructor revisited the **0–100 probability line** drawn in the prior class.
– Students recalled that:
– the left side represented **0**
– the right side represented **100**
– the line itself represented **probability**
– The instructor then asked what the **first dot** placed on the line represented.
– After some discussion, the class identified that the first dot was the **baseline**.
– A student discussion recalled that, in the U.S.–Iran nuclear example, the class had placed the initial estimate at roughly **5%** because:
– nuclear weapons **have been used before**, so the probability is **not 0**
– but their use is **extraordinarily rare**
– the **costs are extremely high**
– and therefore the starting likelihood remains **very low**
– The instructor formalized this as **baseline construction**:
– begin with **historical frequency**
– consider whether similar events have happened before
– consider the **general structural costs** involved
– use that as the **starting point**, not the final answer
## 4) Review of what moves the probability estimate
– The instructor then asked what affects the **final dot** after the baseline is established.
– The class identified **signals** as the major factor.
– The instructor clarified directional movement:
– movement **left** = **decreasing likelihood**
– movement **right** = **increasing likelihood**
– Students and instructor together reviewed:
– **prohibitive costs** move the estimate **left**
– **costly signals** move the estimate **right**
– This was framed as the main analytical task from Thursday:
**establish a baseline, identify costs and signals, then adjust the estimate accordingly.**
## 5) Limits of precision: no universal formula for every forecast
– Before returning to group work, the instructor addressed an important limitation:
there is **no single universal formula** for weighing all costs and signals across all cases.
– He explained that:
– real-world governments and private analytical organizations often use **their own internal weighting systems**
– for this class, students will use a **simplified, partly arbitrary weighting model**
– the point is not to achieve a perfect scientific formula, but to learn how to **reason systematically and defend weights analytically**
## 6) Transition back into group work: reviewing outcomes, signals, and costs
– Students were told to return to the **probability line they had already drawn** and prepare to explain:
– the **outcome** they were forecasting
– the **signals** they had identified
– the **costs** that could inhibit the outcome
– During this transition, a student asked whether they could use **imagined/hypothetical signals**.
– The instructor clarified that the important issue is not whether the signal is “really true” underneath, but **how the signal is interpreted by the other actor**.
– Example: if **Iran says it has a nuclear bomb**, the analytical question is not simply whether Iran truly has one, but **whether the United States believes the claim**
– Another question concerned quantity: whether outcomes must have many signals.
– The instructor answered that for a final paper, students should be **as exhaustive as possible**
– but for the current in-class exercise, **3–4 signals** was sufficient
## 7) Group reporting: Case 1 — U.S. nuclear strike on Iran
– **Fontan Hermine Tiphaine Bagdad** presented for her group.
– Her group kept the same core outcome used in class discussion: **the United States nuking Iran**.
– Signals identified by Hermine’s group included:
– **Iran making it clear that it has nuclear weapons and may use them**
– **the United States publicly announcing that it would retaliate**
– an additional point about a **window of opportunity** in which U.S. decision-makers might believe such an action could “work” at acceptable cost
– the transcript is partially unclear here, but the group was clearly trying to identify a scenario in which the U.S. perceives the strike as more operationally feasible
– Hermine’s group also noted a possible **de-escalatory/prohibitive condition**:
– if **Iran accepted U.S. conditions in negotiations**, that could reduce the probability of escalation
– The instructor added a major prohibitive cost to this case:
– **World War III / nuclear Armageddon**
– He noted that this was a useful extreme example because it tested the outer limits of the model.
## 8) Group reporting: Case 2 — U.S. takes control of Greenland
– A second group presented a Greenland case.
– Initially the group described the outcome as “invading Greenland,” and the instructor pushed them to make the outcome **more precise**.
– Together they refined the forecast into something more measurable:
– **the United States will take control of Greenland**
– The group’s proposed signals included:
– **reopening the consulate in Greenland**
– **increased U.S. resource extraction from Greenland**
– The instructor pressed the group to think carefully about the empirical basis of the resource-extraction claim, noting uncertainty about what extraction is currently taking place.
– The group’s prohibitive costs included:
– **war** if Denmark fought back
– damage to **U.S. international standing**
– a **NATO-on-NATO conflict**
– The instructor elaborated that such a move could seriously damage the alliance and create major diplomatic fallout.
## 9) Group reporting: Case 3 — Chinese blockade of Taiwan
– A third group presented a China/Taiwan case.
– Their outcome was clearly framed as:
– **the likelihood of a Chinese blockade of Taiwan within 18 months**
– The instructor praised the group for including a **timeline**, and used this moment to remind the whole class that **all forecasts must be bounded in time**.
– He recommended a general scope of about **1 year to 1.5 years**
– Longer windows such as **5 or 10 years** would be much harder to model analytically
– The China/Taiwan group proposed signals such as:
– **Xi Jinping publicly announcing plans or intent consistent with a blockade**
– a significant **increase in Chinese naval vessels around Taiwan**
– One student in that group referenced a recent increase in vessel activity and suggested that **further sharp escalation in deployments** would be a strong signal.
– Proposed prohibitive costs included:
– **U.S./NATO tariff escalation**
– broader **economic and business losses**
– damage to **China’s international position**
## 10) Clarifying forecast scope: timeline as a necessary part of the outcome
– **Calmettes Zoe** asked a clarifying question about whether the “timeline” meant:
– the time boundary for the forecast,
– or something to be plotted directly on the probability line
– The instructor clarified that students should primarily set an **endpoint / forecasting window**, such as:
– from now until a date about a year to a year and a half away
– This was an important methodological point:
the class reinforced that **an outcome without a time horizon is too vague to forecast well**.
## 11) Mini-lecture: the problem of weighting costs
– The instructor then shifted from identifying signals/costs to **weighting** them.
– He began with a philosophical question about the **theoretical limit to costs** in international relations.
– One **uncertain student** responded that an actor cannot invest more soldiers than it actually has.
– The instructor used this to explain that, in practice, **costs can approach the effectively infinite**, which makes formal standardization difficult.
– Because cases vary too widely, he decided:
– students will **assign cost values themselves**
– they do **not** need to defend exact percentage-point precision
– but they **do** need to justify the general logic of why a cost drags the probability estimate leftward
## 12) Mini-lecture: building a three-part signal taxonomy
The instructor next introduced a more structured way to evaluate **signals**.
### A. Cheap talk
– The class identified **cheap talk** as the **least valuable signal**.
– Students defined it as language that is **“just words”**.
– The instructor clarified that cheap talk matters only a little because:
– there is **no real cost to lying**
– there is **little or no cost to reversal**
– Initial weighting was discussed as **3–5%**, but after a student challenge, the instructor revised this to a broader and more flexible:
– **1–5%**
– One student asked whether repeated statements should add another 3–5% each time.
– The instructor answered **no**:
– repeated rhetoric should usually be treated as a **single category of signal**
– otherwise the model would become distorted by counting every repeated threat separately
### B. Historical frequency belongs to the baseline, not the signal category
– Another **uncertain student** raised the issue that repeated historical behavior can matter enormously even if it is not a signal in the moment.
– The example given was a case like **China blockading or attempting something multiple times**
– The instructor confirmed that this kind of evidence belongs to the **baseline**, not to signaling.
– This helped reinforce the distinction between:
– **historical precedent** = baseline
– **current cues/messages/actions** = signals
### C. Costly signals
– The instructor then asked for the **most valuable signal category**, leading to discussion of **costly signals**.
– A student suggested an action-based example:
**the United States sending military troops to Greenland**
– The instructor noted that this would be an extremely powerful signal, close to a **sunk cost**, and in some cases almost part of the outcome itself.
– He clarified that signals do **not** have to be verbal:
– a signal may be **speech**
– but it may also be an **action**
– What matters most is **the cost of reversal**:
– if an actor backs down, what political, reputational, or strategic cost does it suffer?
– The instructor used the China/Taiwan example at length:
– if China sends more ships near Taiwan and then one is attacked or challenged,
– backing down would impose **domestic political costs**
– the regime could appear weak
– state media/public opinion could punish leadership for failing to respond
– A student added that reversal could also impose:
– **credibility costs**
– **financial costs**
– increased **retaliation risks**
– The instructor accepted the logic but steered the class toward focusing above all on the **political cost of reversal**, rather than turning the exercise into a narrow financial accounting problem.
– The class then fixed the working range for **costly signals** at:
– **15–25%**
### D. Reversible signals
– The instructor created a middle category between cheap talk and costly signaling:
– **reversible signals**
– These were defined as signals with:
– a **real cost to reversal**
– but **not a catastrophic one**
– Example logic:
– some backlash
– perhaps some poll loss or criticism
– but not career-ending or regime-shaking consequences
– The class assigned a range of:
– **6–14%**
– This completed the three-part signal model:
– **Cheap talk:** 1–5%
– **Reversible:** 6–14%
– **Costly:** 15–25%
## 13) Applied clarification: is a ceasefire a reversible signal?
– An **uncertain student (“Amin” in the transcript; no confident roster match)** asked whether a **ceasefire** should count as a reversible signal.
– The instructor unpacked this by asking what it would mean to **reverse** a ceasefire:
– the agreement breaks down
– hostilities resume
– The discussion then focused on the U.S. perspective in a U.S.–Iran-style scenario.
– The instructor concluded that a ceasefire generally fits the **reversible** category because:
– it has some political cost if broken
– but ceasefires are also understood to be **temporary and fragile**
– breaking one usually does **not** create the same level of binding commitment as a major costly signal
– This exchange helped students operationalize the middle category with a concrete example.
## 14) Applied clarification: why nuclear outcomes remain unlikely even after costly signals
– **Fontan Hermine Tiphaine Bagdad** raised an important conceptual objection:
– even if both sides issue strong costly signals in a nuclear context,
– **nuclear use can still remain very unlikely**
– the Cold War provided a historical example of prolonged, intense signaling without nuclear war
– The instructor agreed and used this to explain the interaction between **signals** and **prohibitive costs**:
– in nuclear scenarios, the **prohibitive costs are so large** that they can overwhelm even strong signals
– thus the baseline may remain low and the final estimate may stay low despite visible escalation
– This was a key synthesis point in the lesson:
**signals do not automatically override structural costs**.
## 15) Final group task: categorize existing signals by reversal cost
– After the lecture on weighting, the instructor sent students back to their groups briefly.
– They were told to take the signals they had already found and classify each one as:
– **cheap talk**
– **reversible**
– **costly**
– The analytical test students were to apply was:
– **What would it mean to reverse this signal?**
– **What would the political cost of that reversal be?**
– The instructor emphasized again that this classification is not about absolute truth, but about **credibility and consequences**.
## 16) Transition to end-of-class administration: presentation briefing
With about 12 minutes remaining, the instructor shifted to logistics and expectations for **next week’s presentations**.
### A. Presentation length and preparation
– Each presentation will be **5 minutes**
– The instructor strongly recommended **rehearsing in advance**
– He specifically warned students not to let the presentation day be the **first time** they actually try delivering it
– He advised practicing:
– in front of a phone camera
– in front of friends/family
– with an actual timer
### B. Slides
– **PowerPoint slides are optional but recommended**
– The instructor warned that slides should remain **limited and supportive**
– He explicitly said that reading from:
– a full script
– or directly from the slides
would significantly reduce the grade
– Notes are acceptable only as **prompts**, not as a word-for-word text to read aloud
### C. Professional delivery and audience role-play
– The instructor asked students to imagine that they are not simply talking to classmates, but to a **decision-maker / boss**
– Example audiences included:
– a boss at a **think tank**
– an **ambassador**
– a member of a **diplomatic corps**
– One student asked whether they should choose a policy based on the identity of the audience.
– The instructor replied that students do **not** need to define a full policy position, but they **should imagine a realistic audience that would care about the forecast**
– He emphasized:
– professional decorum
– respectful address
– speaking as if the audience were a superior in a real workplace
– doing nothing in the presentation that would be inappropriate in front of one’s actual boss
## 17) Required presentation structure: memo format
The instructor laid out the structure students should follow.
### A. BLUF
– The first section should be **BLUF**: **Bottom Line Up Front**
– The BLUF should contain:
– the **most recent update**
– the **main conclusion**
– Students were told that the audience should know almost immediately:
– what has recently happened
– and **what probability judgment the student has reached**
### B. Core facts / ground truth
– The next section should summarize the **key facts** of the case
– This includes:
– the most important developments from the last 1–3 years
– the essential factual background behind the prediction
### C. Analysis
– Students should then explain **why the current situation is what it is**
– The instructor linked this to the kind of reasoning students had already practiced in earlier assignments, especially the midterm
### D. Forecast
– Here students must show **how they arrived at the probability estimate**
– They should discuss:
– the **baseline**
– the **signals**
– the **costs**
– and the logic/equation used to reach the final estimate
### E. Indicators and warnings
– Finally, students should end with **indicators and warnings**
– These are the **conditions that would need to be met** to significantly raise the probability of the forecasted outcome
– This tied directly back to the opening conceptual review of the class
## 18) Final logistical questions and deadlines
– The instructor noted that **Thursday’s class** will be the final real working session before presentations and will be used to:
– construct memos
– finalize scales
– put finishing touches on presentation logic
– Students asked when presentation materials are due.
– The instructor clarified:
– **presentation materials are due Sunday before the presentation**
– if using **PowerPoint**, students should send the slides by Sunday
– if **not** using PowerPoint, students should submit a **very short outline** with bullet points
– He said he would place an **assignment upload sheet on the course site**
– The transcript suggests a Sunday time was mentioned, but the exact time is not fully clear from the recording; the day itself (**Sunday**) was clearly emphasized
– A student asked whether there would be feedback on presentations.
– The instructor said he could **not guarantee individual feedback for everyone during class time**
– He also mentioned that he had finished grading the prior paper and expected students to receive that feedback **by the start of next week**
—
# Student Tracker
– **Calmettes Zoe** — Asked clarifying questions about how to define the forecast timeline and how to classify a ceasefire within the signal framework, helping refine the methodological model.
– **Fontan Hermine Tiphaine Bagdad** — Presented her group’s U.S.–Iran nuclear escalation case and later raised a strong analytical objection about why nuclear outcomes can remain unlikely even when costly signals exist.
– **Uncertain student (name not captured)** — Helped distinguish between asking for necessary conditions and asking under what conditions an outcome would occur.
– **Uncertain student (name not captured)** — Contributed to recalling the baseline discussion around the low probability of U.S. nuclear use against Iran.
– **Uncertain student (Greenland group speaker; name not captured)** — Refined the Greenland case into a measurable outcome and proposed signals such as consular reopening and resource extraction, along with costs such as war and NATO damage.
– **Uncertain student (China/Taiwan group speaker; name not captured)** — Framed the Taiwan case as a blockade forecast within 18 months and identified naval buildup and leadership statements as key signals.
– **Uncertain student (name not captured)** — Noted that repeated historical behavior should matter in probability judgments, prompting clarification that this belongs to the baseline rather than to signals.
– **Uncertain student (name not captured)** — Raised the point that total available soldiers can limit how costs are conceptualized in war.
– **Uncertain student (“Amin” in transcript; could not be confidently matched to the roster)** — Asked whether a ceasefire should be considered a reversible signal.
– **Uncertain student (name not captured)** — Asked whether presentation style should vary according to the kind of policy audience being imagined.
—
# Actionable Items
## High urgency
– **Prepare 5-minute presentation** for next week.
– **Submit presentation materials by Sunday**:
– PowerPoint slides if using slides
– brief outline if not using slides
– **Rehearse presentation aloud** in advance and time it.
– **Avoid reading from slides or a full script**.
## Before Thursday’s class
– Finalize each group’s:
– **bounded outcome**
– **timeline**
– **baseline**
– **signal categories**
– **cost logic**
– Be ready to use Thursday’s class to finish the memo/presentation structure.
## Presentation quality expectations
– Use **professional delivery** and address the audience as if presenting to a boss or decision-maker.
– Organize presentation in this order:
– **BLUF**
– **core facts / ground truth**
– **analysis**
– **forecast**
– **indicators and warnings**
## Instructor follow-up
– **Return prior paper feedback/grades** by the start of next week.
– **Confirm the exact Sunday submission time on the course site**, since the transcript clearly gives the day but the precise hour is less clear.
Homework Instructions:
ASSIGNMENT #1: Forecast Presentation
You will prepare a short professional presentation based on the forecasting model we practiced in class. Your goal is to present your forecast clearly and persuasively by showing the baseline, the signals, the costs, your final probability judgment, and the conditions that would need to be met for the outcome to become more likely.
Instructions:
1. Choose the forecasting outcome you and your group are analyzing.
– Use the same type of outcome discussed in class: a specific event that may or may not happen.
– Make sure your outcome is concrete and clearly worded.
– Avoid vague phrasing. For example, define exactly what counts as the event happening.
2. Set a clear time horizon for your forecast.
– Do not leave the forecast open-ended.
– Bound the outcome in time, as discussed in class, by choosing a reasonable near-term window, generally about 1 year to 1.5 years.
– State that time frame explicitly in your presentation.
3. Re-establish your baseline probability.
– Begin by asking how often this kind of event has happened before or how historically rare/common it is.
– Use that historical context to place your initial “dot” on the probability line before adding any new signals or costs.
– Be ready to explain why your baseline starts where it does.
4. Identify the main signals that affect the probability of the event.
– Review the signals you discussed in class and decide which ones are most relevant to your case.
– For each signal, ask:
– What does this signal communicate?
– What would it mean if the actor reversed course or failed to follow through?
– What would the political or strategic cost of that reversal be?
5. Classify each signal into one of the three categories from class.
– Cheap talk: low or no real cost to reversal; generally worth about 1–5%.
– Reversible signal: some real cost to reversal, but not catastrophic; generally worth about 6–14%.
– Costly signal: high cost to reversal and therefore much more meaningful; generally worth about 15–25%.
– Use your judgment to place each signal in the best category and be prepared to justify that choice.
6. Identify the prohibitive costs that push the probability downward.
– These are the major reasons the event may not occur.
– Think about the kinds of costs discussed in class, such as war escalation, domestic political backlash, damage to alliances, international standing, or other serious consequences.
– You do not need to force these into a strict percentage formula, but you should explain how and why they weigh heavily against the event occurring.
7. Decide on your final forecast.
– After considering the baseline, signals, and costs, determine your final probability estimate.
– This should be the conclusion you present to the audience.
– Make sure your final estimate reflects the overall balance of evidence, not just one dramatic signal.
8. Identify your indicators and warnings.
– List the conditions that would need to be met for the event to become much more likely.
– These are the “under what conditions would it happen?” points emphasized in class.
– Focus on the developments a decision-maker should watch for going forward.
9. Organize your presentation using the required structure from class.
– Your five-minute presentation should follow this order:
1. BLUF (Bottom Line Up Front)
– Start with the most recent update relevant to your case.
– State your main conclusion immediately.
– Tell the audience the final probability judgment up front.
2. Core facts / ground truth
– Summarize the key factual background that led to this forecast.
3. Analysis
– Explain why the current situation looks the way it does.
4. Forecast
– Show how you used the baseline, signals, and costs to reach your conclusion.
5. Indicators and warnings
– Explain what future developments would increase the likelihood of the event.
10. Keep the presentation to five minutes total.
– Your presentation must be short, direct, and disciplined.
– Practice enough so that you can finish within the time limit without rushing or reading too slowly.
11. Practice the presentation before class.
– Do at least one full dry run before the actual presentation day.
– Time yourself while speaking.
– Practice in front of a camera, a friend, or anyone available so that the first time you give it is not during class.
12. Use slides only if they help you.
– Slides are optional but recommended.
– If you use slides, keep them simple and limited.
– Do not fill them with text and do not read from them.
13. Do not write or read a full script during the presentation.
– You may use notes to stay on track, but they should only guide you.
– You should not read word-for-word from paper or from the screen.
– Know your material well enough to explain it professionally in your own words.
14. Present as if you are briefing a real decision-maker.
– Imagine you are delivering this forecast to a boss, diplomat, embassy official, ministry analyst, or think-tank supervisor.
– Keep your tone formal and professional.
– Address the audience respectfully and present yourself as someone giving serious policy analysis.
15. Be prepared to present professionally next week.
– Dress and speak in a professional manner.
– Remember that the assignment is not just about your conclusion, but also about how clearly and credibly you deliver it.
16. Submit your materials by Sunday night.
– If you are using PowerPoint slides, submit the slides by Sunday.
– If you are not using slides, submit a very short outline instead.
– Your outline should include brief bullet points showing the main parts of what you plan to say.
17. If you submit an outline instead of slides, keep it concise.
– A simple 1–5 bullet outline is enough, as long as it gives a clear sense of your presentation plan.
18. Make sure your submitted materials match what you will actually present.
– Your slides or outline should reflect the same forecast, structure, and main argument you plan to deliver in class next week.