Considering that there have been fewer than expected launch failures in recent years compared to the base rate, I adjust up my forecast from that produced by my base rate analysis

## 0.015

##### Relative Brier Score

## 162

##### Forecasts

## 45

##### Upvotes

##### Forecasting Calendar

Past Week | Past Month | Past Year | This Season | All Time | |
---|---|---|---|---|---|

Forecasts | 0 | 24 | 258 | 162 | 311 |

Comments | 0 | 0 | 6 | 4 | 15 |

Questions Forecasted | 0 | 22 | 61 | 51 | 65 |

Upvotes on Comments By This User | 0 | 20 | 51 | 45 | 53 |

Definitions |

#### Active Forecaster

#### Upvotes Received

Updating after correcting a data error in my initial forecast which propagated through:

https://www.infer-pub.com/comments/97509

(have edited, so the rationale is correct now)

Data I used:

2020: 1340 total incidents

2021: 1091 incidents

2022: 625 incidents

2023 (up to July 2nd): 272 incidents

To model the decreasing trend since 2020, I took a logarithmic transformation of the incident counts and applied linear regression to the output value, in order to capture what seems to be something like an exponential decay curve.

I then used a Monte Carlo simulation. I calculated the difference between observed and model-predicted values. Using the mean and SD of these differences (residuals), I ran 10,000 simulations, counting how many fell in each bucket, in order to obtain my final probabilities:

- less than 500: 80.33%
- more than 500 and less than 1000: 19.67%
- more than 1000: ~0.0%

Because I lack confidence in this model, I slightly de-extremize my final probabilities when making my forecast.

We have 3 missions that will very likely go ahead in this period identified by the question writers:

- Peregrine Mission 1, launching on Vulcan Centaur VC2

- Nova-C lander, mission IM-1 (will land on the south pole, so will be especially difficult, only 1 out of 3 attempts has been successful)

- Japan's Moon Sniper (lander) - launching on H-IIA 202

In addition, scheduled for the relevant time period we have more landing missions:

- IM-2, using another Nova-C lander (will land on the south pole, so will be especially difficult)

- Chang'e 6 - Long March 5, which is supposedly scheduled to launch in May 2024.

Since missions tend to get delayed often, and some are close to the end date of this question, I will assume there are only 4 attempts, rather than 5.

I count that there have been 193 missions to the moon:

https://en.wikipedia.org/wiki/List_of_missions_to_the_Moon#Missions_by_date

And of these, **41 have been launch failures, and 152 have not been.**

Then, using the Laplace rule of succession I calculate the probability of a launch failure in any given future moon mission to be** 21.54%**

On the previous moon mission question, conditional on successful launch, I had the base rate analysis:

https://www.infer-pub.com/comments/94277

Laplace rule base rates for post-launch successful missions:

Soft landing on south pole =40%(updated for recent 2 attempts)

Luna mission = 61.5%

Zond mission = 58.3%

Soviet mission = 62.5% (not sure that I got all the data for this calculation)

Any mission =68%(relatively low confidence estimate, but this is what I will be using here for non-south-pole missions)

I will label our 4 missions as:

Mission A - South Pole Landing

Mission B - South Pole Landing

Mission C - Non-South Pole Landing

Mission D - Non-South Pole Landing

For missions landing on the south pole:

P(Launch Successful)=1-0.2154=0.7846

P(Landing Successful | Launch Successful)=0.4

Therefore:

P(Mission **A or B Successful**) = P(Launch Successful)*P(Landing Successful | Launch Successful)=0.7846*0.4=**0.31384**

For missions not landing on the south pole:

P(Launch Successful)=0.7626

P(Landing Successful | Launch Successful)=0.68

Therefore,

P(Mission **C or D Successful**)=0.7846*0.68=**0.53353**

Now, we calculate the probability that no mission is successful:

P(Mission A Fails)=1-0.31384=0.68616

P(Mission B Fails)=0.68616

P(Mission C Fails)=1-0.53353=0.46647

P(Mission D Fails)=0.46647

Therefore:**P(All Fail)**=P(A Fails)*P(B Fails)*P(C Fails)*P(D Fails)

=0.68616*0.68616*0.46647*0.46647

=**0.11020****And hence, our overall probability that at least one moon landing resolves successfully per the question requirements is:**

1-0.110=**89.0%**Repeating the analysis, if we exclude IM-2, we get

**85.3%**

If we include IM-2 and also Chang'e 6, we get **93.1%**

No news on this, still don't see it happening anytime in the near future with ExxonMobil having recently pulled out of this area

Didn't happen

Still don't see this happening in a way that meets the resolution criteria

I put most of my weight towards the higher buckets due to a general increasing trend, presumably driven by increased uptake of autonomous vehicles. There also seems to be seasonality in the data, but would need significantly more time to do a more serious forecast.