What are you currently optimizing that should not exist? Elon's Algorithm: Delete. "The best part is no part. The best process is no process." Knudstorp did not optimize LEGO's theme parks. He sold them. All four Legoland parks. Gone. He killed the clothing line and the video game division the same way. Cut brick types from over 12,000 to roughly 6,000. Cut 1,000 jobs. That was 2004. LEGO was losing $1 million a day. Ask any busy exectuve what they'd cut. They'll rattle off a list. Ask why they haven't. Silence. At LEGO, every day without a decision burned another million. Thirty million a month, compounding into an $800 million hole. The risk was never cutting too deep. The risk was the daily cost of keeping things that shouldn't exist. Elon's test for Step 2: if you don't end up adding back at least 10% of what you removed, you didn't remove enough. Delete until something breaks. Then add that thing back and leave the rest dead. Deletion is not just a strategy problem. It's a psychology problem. Everyone at LEGO knew the theme parks were bleeding cash. Everyone could see 12,000 brick types was insane. They had sets with micro-motors that cost more to produce than their retail price. Every unit sold lost money. And someone's job was to make sure those units shipped on time. Nobody acted because deletion is an indictment. Kill a product line, and you're saying it shouldn't have existed. If you championed it two years ago, that deletion has your name on it. So instead of cutting, people "improve." They reorganize. They study it for another quarter. Anything except the one thing that works. Elon calls this the most common mistake smart engineers make: optimizing a thing that should not exist. Think about what killing one brick type actually eliminates. A mold to maintain. Inventory to track. A production slot. QC steps. Warehouse space. An SKU in the system. Worker training. Packaging variations. One brick is never just one brick. It drags an entire web of downstream complexity behind it. Kill the brick, the web disappears. The things you can't bring yourself to delete aren't the ones you know are useless. They're the ones you've already invested in improving. You renovated the process. Hired a specialist. Bought the software. That investment doesn't make the thing more valuable. It just makes the deletion more painful. Knudstorp didn't ask "what should we fix?" He asked "what should not exist?" The first question gets you a better process. The second kills the process. He didn't add a single thing. Operating margin went from 2.4% to 15.6% in three years. Revenue started growing 15% annually. By 2014, LEGO passed Mattel as the world's largest toy company. The questions for Step 2: If we removed this entirely, what specifically breaks? What would a new owner kill on day one? What are you currently optimizing that should not exist?