diff --git a/functions/onprem/orborus/orborus.go b/functions/onprem/orborus/orborus.go
index a6435b65..312ee7f1 100755
--- a/functions/onprem/orborus/orborus.go
+++ b/functions/onprem/orborus/orborus.go
@@ -2183,7 +2183,6 @@ func main() {
 		fullUrl += "?amount=50"
 	}
 
-
 	if isKubernetes == "true" {
 		log.Printf("[INFO] Finished configuring kubernetes environment. Connecting to %s", fullUrl)
 	} else {
@@ -2532,6 +2531,8 @@ func main() {
 				executionRequests.Data = executionRequests.Data[0:allowed]
 			}
 		} else if swarmControlMode && (swarmConfig == "run" || swarmConfig == "swarm") {
+			// any reason it is not maxConcurrency instead of 
+			// hardcoded 50?
 			if len(executionRequests.Data) > 50 {
 				executionRequests.Data = executionRequests.Data[0:50]
 			}
@@ -4060,6 +4061,10 @@ func sendWorkerRequest(workflowExecution shuffle.ExecutionRequest, image string,
 	return nil
 }
 
+// 0x0elliot:
+// let's never increase worker replicas.
+// in our tests, workers replicas mattered a lot less.
+// edge-case: subflows are helped with when worker replicas are higher.
 func AutoScale(ctx context.Context) {
 	if os.Getenv("SHUFFLE_SCALE_REPLICAS") != "" {
 		return
diff --git a/functions/onprem/worker/worker.go b/functions/onprem/worker/worker.go
index ea3b748d..567b52b8 100644
--- a/functions/onprem/worker/worker.go
+++ b/functions/onprem/worker/worker.go
@@ -4806,6 +4806,12 @@ func runWebserver(listener net.Listener) {
 	}
 }
 
+// 0x0elliot:
+// IF we had to rewrite this, we will focus on ONLY auto scale for apps.
+// i recommend we target executions/minute (?) as a metric.
+// edge-case: subflows are helped with when worker replicas are higher.
+// i kind of never want to scale down. at least, not now.
+// also, algorithm is very broken. executions/worker
 func AutoScaleApps(ctx context.Context, client *dockerclient.Client, maxExecutionsPerMinute int) {
 	ticker := time.NewTicker(1 * time.Second)