35,848 results · Most relevant
Tom Foley outlines a valuable process for young type designers: take a historical typeface model, photograph it, analyse it, rationalise it, and then develop digital drawings from it. He notes that reviving a classic typeface (such as Caslon, Bodoni, or Baskerville) teaches proportion, shape, form, and contrast — the vernacular of a type designer.
Gary Vaynerchuk describes his '1585' system: he wants his team to involve him in the first 15% of everything he's involved in — because he needs to ideate and contribute his strategic and intuition feeling — then he lets them completely own the next 80%, and he comes in at the last 5% to put a cherry on top. He says it really works for him and considers it the greatest system of his career, in the prime of it right now, and he's thinking about writing a book about it called '1585'.
attributed_claim
The speaker recommends starting to make videos on the internet about what you're good at — salesmanship, being a good partner when you're busy, building up managers, how to properly sell, or how to create lifetime value rather than a one-off sale. If you don't like being on video, write; if video feels cringe, do a podcast and just do audio. He notes that being silent on the internet is a very bad strategic move.
asserted
The host recommends stacking 'following your company' + 'posted on LinkedIn' + ICP job titles in Sales Navigator to surface warm, responsive prospects, claiming the three main filters can book a meeting within 24 hours.
speaker's promise, not demonstrated
Seinfeld's production week ran: day one table read and light rehearsal (actors knew material would change); overnight rewrites by Larry and Jerry; day three full rewrite overnight plus a run-through for NBC and Castle Rock whose notes Larry David would push back on and diminish; day four camera block day when cameramen learned shots and assignments were finalized; day five morning rehearsal then filming with the live audience in the afternoon, running the show in chronological order.
Use a tool like Playwright to check that an email was actually saved when it gets on the waitlist. Check that the waitlist has no emails, fill in an email, click the join waitlist button, then expect the status element to contain 'waitlisted' and verify the email address is in the inner text. Give end-to-end test tools like Playwright or React Testing Library to your agent as a tool so it can prove things out itself.
One technique for managing context relevance across subagents is to gradually reveal context by having a markdown file that links to other markdown files: the main document acts as an index (e.g., for an engineering team working on the kernel: kernel here, assembler here, editor here) so subagents read only the details relevant to them, saving tokens.
endorsed (technique), hypothetical (example)
Dr. MacDougall describes the modern approach to deep brain tumors: a two-millimeter drill hole in the skull through which a fiber optic cannula attached to a laser heats the tumor under real-time MRI visualization. MRI pictures are taken every second or so as the tumor heats up, monitoring temperature so it kills tumor cells without hurting surrounding brain. Tumors previously considered totally inoperable — such as in the brainstem, primary motor cortex, or Broca's area — can now sometimes be removed with patients coming out unscathed.
Consequential creator advice requires independent verification before Today can promote it.
Harish Gaggar describes building MCP gateway monitoring that sends all operation logs from agent calls into a log store (ELK or a cloud virtual server), which can be queried for operational questions like showing all 401s in the last 10 minutes or investigating a failed token refresh. Real-time traces and metrics go to OpenTelemetry, an open source solution that collects, processes, and redirects data to backends: traces go to Tempo (open source trace storage), and metrics go to Prometheus, capturing things like P95 latencies for a given tool and whether auth rejections are spiking at particular times. Aggregated traces and metrics are saved into a graph database for visualizations. Trace data includes which tool was slow on a request, whether the call came from an agent or IDE, and the
Harish Gaggar describes agent memory management as a process to save and restore the state of the workflow over time, ensuring durability and seamless recovery across every query run. When a query is submitted, the agent generates a thread, and the first question on that thread creates a run ID; when a deeper question is asked, the agent goes to a particular node in the LangGraph, executes the query, and saves a checkpoint at that node; follow-up questions on the same thread create another run ID and save checkpoints at specific nodes. For storage, he recommends SQLite DB for local development and Cloud Spanner for production on GCP, noting Azure or AWS can be used as well.
To estimate VO2 max without a lab, run the 12-minute Cooper test for maximum distance (typically one to two-plus miles depending on fitness). As a gentler option — good for people who can't run hard due to knee or back pain — walk one mile submaximally, record your time and ending heart rate, and enter them into a standard calculator; done correctly it is pretty accurate.
several figures flagged possible_numeric_error
To assess muscular endurance, take the load from your prior strength test, set it to 75%, and perform as many repetitions as possible. If you get fewer than 8 repetitions, you have a muscular endurance problem. Because endurance is local, you need to run this test for every muscle group.
the 8-rep threshold carries a possible_numeric_error flag
The cost-free test for power is a simple broad jump: stand in normal position, jump out as far in front of you as possible, and measure the distance between your start point and the back of your heel where you land. Benchmark: you should be able to broad jump at least your own height (e.g., five foot five for someone 5'5). For females, ratchet the benchmark down about 15%, because women in general don't have the power men have. Galpin notes this is a very crude number.
benchmark described as crude
A three-day split is probably the best way to complete the full testing battery, ideally at the start of the year or whenever you change your training. Do body composition tests (FFMI, body fat) first, after at least 48 hours of rest from hard exercise, since acute exercise skews measurements via inflammation and water storage. Do movement assessments when freshest. Skill, max strength, and power tests go at the very beginning of the day; fatiguing tests at the end. On one day you can roll from the broad jump into leg strength, then leg muscular endurance while warm — e.g., a 1RM leg extension, rest five to seven minutes, load to 75%, and do max reps. The anaerobic capacity test goes on a separate day or after the long-duration test (which acts as a big warm-up), and the VO2 max test gets,
To assess anaerobic capacity, you can go to a laboratory and do a Wingate test — a 30-second maximal test measuring how much work you can do in 30 seconds. In the lab, a Bosco protocol is also often used: stand on a force plate and do as many vertical jumps as fast as you can until destroyed around second 45. Without lab access, use any maximal-effort protocol such as sprinting, an air bike, or a rower for roughly 30 seconds up to 45 seconds or a minute. Choose a modality where you can exert maximal effort without technical problems — avoid technically complex movements like kettlebell swings because there are too many other variables. Go as hard as you possibly can, knowing you will reach tremendous fatigue. There are no real standards because distance covered varies too much by modality;
no real standards; modality-dependent
For long duration steady state exercise, the standard is to maintain consistent work output for over 20-plus minutes (minimal target generally 20 minutes, more typical range 20 to 60 minutes). Pick something already in your lifestyle, like a loop around your house, and test your ability to maintain work without stopping. A personal twist: do it with nasal breathing only. Going 30 straight minutes without a break is a good benchmark, but walking doesn't really cut it unless you're very unfit — if walking 30 minutes without a break is a challenge for you, that's an acceptable starting point. If you can, move at a non-walking pace; the zone (2, 3, 4, or 5) doesn't matter. Show you can maintain a minimum of 20 minutes of work with no breaks, no intervals, no downtime, and ideally breathing
benchmarks presented as general guides
To assess muscular endurance, take the exact strength test you did earlier, load it to 75%, and do as many repetitions as you can. For example, if you could do 200 pounds on your leg extension test, put 75% of that load on and do as many reps as possible. You want to look for more than eight repetitions — being below eight repetitions at 75% of your max load indicates a muscular endurance problem.
threshold presented as a general guide
Without a lab, do a 30-second maximal effort on any modality allowing max effort without technical problems (sprinting, air bike, rower). In the lab, a Bosco protocol is used: repeated maximal vertical jumps on a force plate until exhaustion around second 45. The key self-check is reaching close to predicted maximum heart rate (220 minus age — an extremely generic number). The better metric is heart rate recovery: after maximum exhaustion, roughly half a beat per second — about a 30 bpm drop within 60 seconds, 60 beats within two minutes, and half again by three minutes. Worse recovery indicates a problem in anaerobic or cardiovascular capacity.
220-minus-age is explicitly called an extremely generic number; some figures flagged possible_numeric_error
To test heart rate recovery, get up to a place of absolute terrible exhaustion and maximum fatigue, then test your heart rate with a timer going and count. Good heart rate recovery is about half a beat per second: roughly 30 beats per minute drop within 60 seconds. By two minutes, your heart rate should have dropped by 60 beats total; by three minutes, roughly 90 beats total. If your recovery is worse than these benchmarks, you have a problem in anaerobic capacity or cardiovascular capacity.
benchmarks presented as general guides
Eugene Teo recommends having some starting metrics you can check in with every few weeks: strength numbers going up as benchmarks for strength work, and athletic-style metrics like broad jump, triple broad jump, hops, and vertical jump. You might not feel the difference day to day or week to week, but it's nice to have something to look back on like a progress picture to see what you've worked towards. There doesn't always have to be a strict metric like adding 10 inches to your vertical — create some kind of baseline for where you're at and give yourself as many different things as possible to grab onto to look for improvement in the future.
Choose sources → to get a library of your own and a daily brief.