**Lex Fridman** (0:00)
The following is a conversation with Irving Finkel, who is a scholar of ancient languages, curator at the British Museum for over 45 years, and is a much admired and respected world expert on cuneiform script, and more generally on ancient languages of Sumerian, Acadian and Babylonian. And also on ancient board games, and Mesopotamian magic, medicine, literature and culture. I should also mention that both on and off the mic, Irving was a super kind and fun person to talk to, with an infectious enthusiasm for ancient history that of course I already love, but fell in love with even more. And now a quick few second mention of each sponsor. Check them out in the description or at lexfridman.com/sponsors. It is in fact the best way to support this podcast. We got Shopify for selling stuff online, Miro for brainstorming ideas with your team, Chevron for reliable energy that powers data centers, Element for electrolytes and AG1 for my daily multivitamin. Choose wisely my friends. And now onto the full ad reads. I do try to make them interesting, but if you skip, please still check out the sponsors. I enjoy their stuff, maybe you will too. To get in touch with me for whatever reason, go to lexfridman.com/contact. All right, let's go. This episode is brought to you by Shopify, a platform designed for anyone to sell anywhere with a great looking online store. So of course Shopify is great as an app, as a service, but the thing that always fascinates me is the engineering behind the scenes. And this is truly an incredible engineering team. And I could talk about this for many hours and perhaps it's good to give an example. Like they built a custom search engine rather than using Elasticsearch or any of the off-the-shelf engines that will require significant rearchitecture at their scale. C++ was the core language for close to hardware optimizations, memory efficiency across hundreds of millions of items. They created rank flow, a domain specific language combining Python-like simplicity with C++ performance, rejected hybrid Python calling C++ approaches due to deployment complexity, version skew and operational overhead. I'm reading these things from their recent blog about how they recommend. There's so many technical blogs about the various problems they're solving. And they do this incremental phase type development deployment where they ship fast, optimize later, maintain compatibility always. So in phase one was the first four weeks, they built SimScore DSL, which is a basic C++ engine to unblock data science experiments. And then in phase two, which took two to four months, they built Turbo DSL, which is a high-performance engine version of that, which achieves 48% speed. There's so much more to talk about, but incredible engineering that brings you an incredible product at the end. Sign up for a $1 per month trial period at shopify.com/lex. That's all lowercase. Go to shopify.com/lex to take your business to the next level today. This episode is also brought to you by Miro, an online collaborative platform. The very kind that the peoples and the civilizations of the ancient world that Irving talks about didn't have. Now, imagine that. One of the great things about the competing age is that we can work together. It's not just the productivity gains for the individuals. It's the productivity gains for teams. And nowhere is that more true than in the process of idea development. As I talked about in a recent episode of Mike 11, do we have ideas or do ideas have us? And ideas have a way of kind of occupying for a time or for a generation, the brains of multiples of people. And those ideas are formed and shaped and modified and evolved across those brains, utilizing those brains. Or the brains do the modification. Whichever it is, you want to have the best tools for the job. And Miro is incredible for that. Converts sticky notes, screenshots, and so on, into diagrams or prototypes in minutes. Super easy to use, makes teamwork fun, help your teams develop great ideas into results with Miro. Go to miro.com to find out how. That's miro.com. This episode is also brought to you by Chevron, an energy company that delivers affordable, reliable energy to US data centers. I'm attending NeurIPS in San Diego, which is a machine learning conference. And boys, one thing clear, aside all the fascinating technical details, the turmoil of the research world, the excitement, the wonder, the mystery, beyond all that is the reality that in order for the scaling laws to hold, the compute needs to scale. And for the compute to scale, we need energy. In the United States, the scaling of the energy infrastructure is essential because the demand for electricity is growing at an unprecedented scale. Chevron is working hard to provide multi-gigawatts of delivered power with the flexibility to scale further. Energy is not an easy problem, especially if the scaling laws hold, especially if there's benefits to the products that rely on artificial intelligence both for the training side and the inference side. And frankly, I think it is the inference side that will over time consume more and more energy and require more and more compute. What a fascinating world we live in. The full stack is full of scientific and engineering challenges. And I love it. Visit chevron.com/power to learn more. That's chevron.com/power. This episode is also brought to you by element, my daily zero sugar and delicious electrolyte mix. Every once in a while, I'll also partake in that sparkling water, but majority of my element consumption is the OG drink mix. I can't live without it. The electrolyte, sodium, potassium, magnesium are essential if you're doing fasting, if you're doing a carnivore, low carb diet, if you're doing anything where your body is pushed to the limit at all, you have to make sure you get the electrolytes right. My favorite flavor is watermelon salt. I had a really nice family Thanksgiving in Ohio.
106 more minutes of transcript below
Try it now — copy, paste, done:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000741051589
Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.
Get 100 transcripts — $10Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000741051589