![](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="250" height="250"><rect fill-opacity="0"/></svg>)
Snaps
2021-10-01
Why We Killed Our End-to-End Test Suite - Building Nubank
When you're building a large system, what you're really building is a high-speed testing and logging system, not the actual architecture itself. Sure wish more people knew this instead of continuing to do it ass-backwards. There's a zillion kinds of testing, but testing is how the org learns, and a fast learning loop has to come ahead of any other consideration (although that loop first exists at the business layer) Here's a random blog essay
Why Science Can't Settle Political Disputes
Taylorism, and its many failures in both political life and industry) is probably the best example of how a physics-based scientific mentality about managing things in the real world seems obvious but is deeply flawed. It's not that it's always wrong. It's that it is intuitively obvious and works until it catastropically fails. It's important to know why
Offline First
HN Discussion on Offline First. Some readers complain they still haven't found the holy architecture for offline first, which to them means replicating the entire backend on the client. No doubt. Lotsa hidden assumptions in the way hackers talk about this. I love the SQLite/S3/Lambda comment by smackeyacky
Why We Killed Our End-to-End Test Suite - Building Nubank
When you're building a large system, what you're really building is a high-speed testing and logging system, not the actual architecture itself. Sure wish more people knew this instead of continuing to do it ass-backwards. There's a zillion kinds of testing, but testing is how the org learns, and a fast learning loop has to come ahead of any other consideration (although that loop first exists at the business layer) Here's a random blog essay
When you're building a large system, what you're really building is a high-speed testing and logging system, not the actual architecture itself. Sure wish more people knew this instead of continuing to do it ass-backwards. There's a zillion kinds of testing, but testing is how the org learns, and a fast learning loop has to come ahead of any other consideration (although that loop first exists at the business layer) Here's a random blog essay
Why Science Can't Settle Political Disputes
Taylorism, and its many failures in both political life and industry) is probably the best example of how a physics-based scientific mentality about managing things in the real world seems obvious but is deeply flawed. It's not that it's always wrong. It's that it is intuitively obvious and works until it catastropically fails. It's important to know why
Taylorism, and its many failures in both political life and industry) is probably the best example of how a physics-based scientific mentality about managing things in the real world seems obvious but is deeply flawed. It's not that it's always wrong. It's that it is intuitively obvious and works until it catastropically fails. It's important to know why
Offline First
HN Discussion on Offline First. Some readers complain they still haven't found the holy architecture for offline first, which to them means replicating the entire backend on the client. No doubt. Lotsa hidden assumptions in the way hackers talk about this. I love the SQLite/S3/Lambda comment by smackeyacky
HN Discussion on Offline First. Some readers complain they still haven't found the holy architecture for offline first, which to them means replicating the entire backend on the client. No doubt. Lotsa hidden assumptions in the way hackers talk about this. I love the SQLite/S3/Lambda comment by smackeyacky