diff --git a/ndc-ldn-2024/readme.md b/ndc-ldn-2024/readme.md
index e42a9ca..c6bf2bc 100644
--- a/ndc-ldn-2024/readme.md
+++ b/ndc-ldn-2024/readme.md
@@ -434,7 +434,7 @@ A: pray 😂 communicate why it’s a problem. Need people to see why it’s a b
 - suggest code changes rather than just saying something is bad / wrong. Teach.
 - framing feedback as questions
 - if your arguing in code reviews, probably doing it wrong.
-- 
+ 
 
 ## How to get a grip on your microservices system using a service-mesh
 
@@ -449,10 +449,55 @@ A: pray 😂 communicate why it’s a problem. Need people to see why it’s a b
 - kiali very cool graph views ok k8s injections and traffic etc between them
 - service mesh - power comes from mirroring requests to new versions of a service. Ie test a new version with real live data (obv any transactions etc have to be separate db etc)
 - chaos engineering, injection faults, including network level style failures
-- 
+
+
 
 ## Let's build the worst Event Sourcing system!
 
+1: 21/34
+
+![worst es](worst-es.jpg)
+
+- project franz 
+- worst es system ever
+- event store db
+- Kafka for event storing or event bridge
+- idempotency
+- event ordering
+- the queue we are standing in is always the slowest 😂
+- Kafka topic and partition split. Topic is logical, partition is physical.
+- event disordering 😂
+- process managers. 
+- the confusion between event sourcing and event streaming
+- optimistic concurrency
+- strong consistency
+- Kafka can’t do the optimist concurrency
+- basically don’t use Kafka for event sourcing.
+- Kafka is good for what it should be used for.
+- Kafka is for event streaming.
+- event sourcing is about recording and storing business events.
+- event streaming is about moving data from one place to another.
+- cap theorem - consistency, availability, partitioning - can’t have all 3
+- outbox pattern, store state and output message at same time. Both or neither
+- CRAP theorem - managers wanting 100% of c an and p…
+- passive aggressive events
+- eg the bin is full, the dishwasher has finished 😂😂😂😂
+- stream design. Event sourcing is still just storage.
+- events on the inside and events on the outside.
+- domain and integration events.
+- important to have the distinction.
+- keep granular events in your domain, in your domains own event store.
+- long living streams. More events a stream, iops goes up
+- snapshots. Every event? Some period of time? Every x events?
+- hope driven design HDD
+- snaps shots are a cache…
+- closing the book. Eg rather than account, accounting month. Stream restarts each month.
+- a bad system will beat a good person every time.
+- fear uncertainty and doubt
+- Peter principle
+- running with an empty barrow
+- 
+
 ## It’s time to rebuild DevOps.
 
 ## CS Fundamentals: Why SSL and SSH are Secure
diff --git a/ndc-ldn-2024/worst-es.jpg b/ndc-ldn-2024/worst-es.jpg
new file mode 100644
index 0000000..8c4d6c4
Binary files /dev/null and b/ndc-ldn-2024/worst-es.jpg differ