Snowball

Podcast 2 — In conversation

Speaker A: The snowball effect is one of those concepts that appears under different names in different fields and yet is immediately recognisable regardless of the terminology. In economics it appears as positive feedback loops and compounding returns. In epidemiology it appears as exponential growth and R numbers greater than one. In social psychology it appears as bandwagon effects and information cascades. In engineering it appears as runaway amplification and thermal runaway. The underlying dynamic is always the same — growth feeding growth, momentum feeding momentum — and the snowball image captures it with a clarity that the technical terms often do not.

Speaker B: The debt snowball method attributed to Dave Ramsey is a good example of a financial strategy that makes psychological sense even when it does not make pure mathematical sense. Mathematically, you should pay off the highest-interest debt first to minimise total interest paid. But the snowball method — smallest debt first — provides the motivational momentum of quick early wins: paying off a small debt completely feels like a tangible victory that keeps you engaged with the repayment process. The mathematical cost of this psychological benefit — paying slightly more interest overall — may be worth it if the alternative is abandoning the repayment effort entirely.

Speaker A: Snowball versus avalanche — as debt repayment strategies?

Speaker B: Both are debt repayment frameworks but with opposite prioritisation. The snowball method — smallest debt first — focuses on psychological momentum and quick wins. The avalanche method — highest interest debt first — focuses on mathematical efficiency and minimising total interest paid. Both involve rolling the freed payment from a cleared debt onto the next debt to accelerate repayment. The avalanche method will, in theory, always cost less overall. The snowball method may be more sustainable for people who need the motivational reinforcement of quick visible progress. The choice between them is a psychological question as much as a financial one.

Speaker A: What is a common mistake about the snowball metaphor?

Speaker B: Assuming it always implies positive or desirable growth. The snowball effect is equally applicable to negative compounding processes — a small misunderstanding that snowballs into a full breakdown of a relationship, a minor technical issue that snowballs into a system-wide failure, a small deficit that snowballs into insolvency. The neutral physics of the snowball — it grows with whatever it rolls through — means that the metaphor is just as available for describing destructive acceleration as constructive growth. Speakers who use snowball to describe desirable momentum should be aware that the same word applies equally to undesirable runaway processes.

Speaker A: Any close synonyms?

Speaker B: Compound, accumulate, escalate, multiply, and spiral all capture parts of the snowball effect's territory. Escalate emphasises increasing severity or intensity, often with a negative connotation. Spiral shares the sense of accelerating growth, particularly in negative contexts — costs spiralling. Compound and accumulate are more neutral and focus on the quantitative growth without necessarily implying acceleration. Mushroom describes sudden rapid growth without the rolling momentum image. None of these captures quite the full snowball picture — the starting small, the self-reinforcing rolling momentum, and the growing scale all in a single word.

Podcast 3 — Prompt engineering

Speaker B: Welcome to the workshop. Today we prompt AI using the word snowball — building the compounding growth systems, viral referral engines, exponential accumulation trackers, feedback loop monitors, and debt snowball calculators that model, track, and manage self-reinforcing growth or repayment processes.

Speaker A: So snowball tells the AI to build the compounding and accelerating systems — the referral engines where each new user brings in more users, the compounding interest calculators, the viral growth trackers, and the debt repayment snowball planners?

Speaker B: Exactly. It signals compounding growth, viral mechanics, and self-reinforcing acceleration. Let me start with a viral referral engine.

Build a viral referral snowball engine: each new user receives a unique referral code. When they refer another user who signs up and completes onboarding, both the referrer and the referee receive a configurable reward. Track each user's referral tree — who they referred, and who those users referred in turn, up to five levels deep. Calculate the viral coefficient — the average number of new users each existing user brings in — and alert the growth team when the viral coefficient drops below one, which means the referral programme is no longer self-sustaining. Display a live referral snowball dashboard.

Speaker B: Viral referral engine issues unique codes, rewards both parties on successful referral completion, tracks five-level referral trees, calculates viral coefficient, alerts when coefficient drops below one, and provides a live referral dashboard — the AI builds the referral tracker, the reward engine, the tree traversal, the coefficient calculator, the alert, and the dashboard.

Speaker A: Alerting when the viral coefficient drops below one is the referral programme health signal: a coefficient above one means the programme is growing on its own — each user brings in more than one additional user on average; a coefficient below one means the user base will shrink unless supplemented by other acquisition channels, and the alert triggers while there is still time to improve the programme.

Speaker B: Viral coefficient below-one alert as the referral programme self-sustaining health signal. Now the compounding schema.

Design a compounding growth tracking schema: a growth_series table recording a named compounding metric — such as user count or revenue or engagement score — with a series ID, name, starting value, and the compounding interval. A growth_datapoints table recording each measurement with the series ID, the period, the measured value, the period-over-period growth rate, and the compounding factor to date. A referral_tree table recording referral relationships with a referrer user ID, a referee user ID, the referral depth level, the referral date, and whether the referral converted. A viral_metrics_view computing the current viral coefficient, the average referral depth, and the total referral tree size per series.

Speaker B: Compounding growth schema tracks named metric series with intervals, periodic datapoints with growth rates and compounding factors, referral tree relationships with depth and conversion, and a viral metrics view with coefficient and tree size — the AI builds the four-table schema.

Speaker A: Recording the compounding factor to date — the cumulative multiple by which the metric has grown from its starting value — alongside the period-over-period growth rate gives both the incremental signal and the long-term magnitude: a series growing at two percent per period may not look dramatic in any single period, but a compounding factor of five-point-two after eighty periods reveals the snowball's scale.

Speaker B: Cumulative compounding factor as the long-term snowball magnitude indicator. Next, the growth visualiser interface.

Design a compounding growth visualiser for a product analytics dashboard: show a line chart of a key metric over time with the option to overlay an exponential trendline based on historical growth rate. Display the current viral coefficient as a gauge with green for above one and red for below one. Show a referral tree visualisation for the top five referrers, expanding downward through the levels of their referral networks. Include a projection panel showing where the metric will be in thirty, sixty, and ninety days at the current growth rate. Allow adjusting the assumed future growth rate to model optimistic and pessimistic scenarios.

Speaker B: Growth visualiser shows a metric chart with exponential trendline overlay, a viral coefficient gauge, a top-five referrer tree expansion, thirty-to-ninety-day projections at current rate, and adjustable growth rate scenario modelling — the AI builds the chart, the trendline, the coefficient gauge, the tree visualisation, and the projection panel.

Speaker A: The adjustable future growth rate for scenario modelling turns the projection panel from a single-point forecast into a scenario explorer: the growth team can see the metric trajectory under their optimistic target rate versus the current observed rate versus a conservative scenario, and the visual gap between the three lines makes the consequence of different growth assumptions immediately apparent.

Speaker B: Adjustable scenario growth rate as the projection assumption visual consequence tool. On to debt snowball.

Create a debt snowball repayment calculator and tracker: accept a list of debts with their current balance, minimum payment, and interest rate. Calculate both the snowball plan — smallest balance first — and the avalanche plan — highest interest first — showing the total interest paid, the total months to debt-free, and a month-by-month repayment schedule for each plan. Allow the user to enter an extra monthly payment amount to add to the current target debt. Show the interest saved versus the minimum-payment-only baseline. Track actual payments made against the plan and recalculate the remaining schedule after each payment.

Speaker B: Debt snowball and avalanche calculator computes both plans with total interest, months to debt-free, and month-by-month schedules, supports extra payment input, shows interest saved versus baseline, and tracks actual payments with schedule recalculation — the AI builds the dual plan calculator, the extra payment handler, the savings comparison, and the actual payment tracker.

Speaker A: Showing the interest saved versus the minimum-payment-only baseline is the motivational anchor that makes the calculator compelling: seeing that committing to an extra one hundred pounds per month saves three thousand pounds in interest and cuts four years off the repayment timeline makes the sacrifice of the extra payment feel concrete and worthwhile.

Speaker B: Interest saved versus baseline as the extra payment motivational anchor. Now something whole.

Build a growth and compounding analytics platform: track multiple compounding metrics — user growth, revenue growth, engagement score — with their viral coefficients and referral trees. Model and compare growth scenarios. Run a debt snowball repayment tracker for users managing multiple liabilities. Provide a unified dashboard showing all compounding processes in one place. Alert users when any metric's growth rate drops below a target threshold. Generate weekly growth briefings with trend analysis and projections.

Speaker B: Growth and compounding platform tracks multiple metrics with viral coefficients and referral trees, models growth scenarios, includes a debt snowball tracker, provides a unified dashboard, alerts on below-target growth rates, and generates weekly trend briefings with projections — the AI builds the full compounding analytics platform.

Speaker A: Tracking multiple compounding processes in a unified dashboard — user growth, revenue, engagement, and debt repayment together — allows the user to see the relationships between their compounding processes: a platform growing its user base should also see compounding revenue growth; if the two are diverging, the dashboard reveals the divergence that would be invisible looking at each metric in isolation.

Speaker B: Unified compounding dashboard as the cross-metric divergence visibility tool. Last one — event snowball tracker.

Write an event cascade and snowball tracker: monitor a system for cascading failure events where one event triggers others, which in turn trigger more — a snowball of failures. Record each event with its cause — whether it was triggered by another event or by an external cause — building a causal chain graph. Calculate the cascade depth — the longest chain of triggered events — and the cascade breadth — the maximum number of simultaneous triggered events at any depth level. Alert when a cascade exceeds a configured depth or breadth threshold. Display the event cascade graph visually in the monitoring dashboard.

Speaker B: Event cascade tracker records each event with its causal trigger, builds a causal chain graph, calculates cascade depth and breadth, alerts on configured depth and breadth thresholds, and displays the cascade graph visually — the AI builds the event recorder, the causal chain grapher, the depth and breadth calculator, the threshold alert, and the visual graph display.

Speaker A: Calculating both cascade depth — the longest causal chain — and cascade breadth — the maximum simultaneous events at any level — gives two complementary measures of cascade severity: a deep narrow cascade may represent a slow-burning sequential failure; a shallow but wide cascade represents a sudden fan-out of simultaneous failures; each pattern requires a different incident response strategy.

Speaker B: That is the goal — one rolling word, better-tracked compounding systems. Keep it in your kit.