I started thinking about this while leading an internal Ops platform at a previous operations-intensive tech company. As we automated Ops work, I found the bottleneck to shifting left was was defining the processes our Ops team followed. Eng entered late: first, we'd throw Ops resources at a new challenge; then gradually systematize it; wait for PM attention; document the process; and finally engage engineering. This was necessary since systematized bad processes meant scaled mishandling of issues and engineering rework.
Drafting AI offers a different path: teams begin automating processes organically as they discover them, growing toward full automation as understanding deepens. Human review catches edge cases, handles ambiguous situations, and gives an incremental path towards full automation without rework.
Hopefully this lets internal tools teams shift automation further left.
I'm looking for something similar: I want it to trigger when I'm on Twitter (or HN) during 9-5 hours, and ask me what I'm supposed to be working on. Just a small nudge to rubber duck what's blocking me
totally— this is for integrating into custom internal tools. Consider content moderation: you wouldn't want mods getting tickets in JIRA then switching to a different tool to action them.
If switching between screens is not a problem, by all means prefer Linear/Asana/Jira/etc!
Totally— OpsQueue is for custom internal React apps to have integrated task management. If workers can switch between Linear and your internal apps, or you don’t have a custom internal React app, you should use Linear.
In some ways OpsQueue’s goal is to be a headless, API-first Linear.
AI will solve web accessibility by screen readers that summarize visual content, ignoring ARIA and making it irrelevant. Multimodal GPT-4 can take a screenshot jpeg and answer questions about what’s in it (buttons, links, ads, headers, etc). The future of accessibility is rendering DOM to jpeg and asking GPT to be your eyes; we’ll look back on semantic markup as a failed idea that was never going to work
Law and Order is great because it's formulaic. You know what you're going to get every time, like a Big Mac in a town you've never been to before. It's not competing with prestige TV, it's competing with rewatching old shows (where you also know what's going to happen).
The punchline: GPT-powered infinite Law and Order might actually outcompete Dick Wolf-powered Law and Order
Prompt: write html using tailwindcss for a landing page of a website called "AI incident reports" that boils down to "there have been 0 incidents of AIs causing harm. The website should look like it's meant to have lists of incidents, graphs, and so on. Consider what someone thoughtful about AI x-risk would want to track. And then have 0 for all the categories and graphs.
Let's think step by step: what should be on the website?
This was a consequence of near-zero interest rates, and will be different now that the FED is moving us to an era of nonzero interest rates.
> Suppose you develop a new technology that is valuable to some industry. The old approach was to sell or license your technology to the existing companies in that industry. The new approach is to build a complete, end-to-end product or service that bypasses existing companies.
Technology (software) is a red herring. Read as “suppose you find some way to make an existing business better”. Might be a better version of a product, cheaper way to manufacture, more effective way to market, etc. How do you monetize that? 2 common choices are:
1. Sell the improvement to existing businesses
2. Make a new business to compete with the existing players, and use your new thing as a competitive advantage
The higher the interest rate, the more effective #1 is; the lower the interest rate, the more effective #2 is.
Technology, which OP cited, is just one way to make a company better. In 2014, it was a very common way, since few people were doing it effectively.
In 2009, the FED started QE (near zero interest rates) in response to the ‘08 financial crisis. Consequently type #2 startups started outperforming type #1 startups. Founders and VCs noticed, and started producing more type #2 startups than type #1 compared to before. It didn’t happen overnight, and this is what cdixon was observing in 2014.
Why are interest rates connected to starting new businesses? Consider Uber entering a new city in 2014. Uber is better than taxis because it has an app (for the sake of argument), but traditional taxis have an advantage of an existing network effect of riders and drivers. To build a comparable network and be competitive, Uber can subsidize drivers to attract riders, until it becomes self sustaining and Uber can switch from subsidizing rides to extracting a profit. At the money level, this looks like Uber spending a large lump sum today to make an ongoing profit stream in the future (when they win the market).
In fact, this is the classic structure of an investment: spend a lump sum today to get a profit stream tomorrow. Same as building a factory. So when is it worth it? When is the profit stream worth spending a lump sum on, and when is the lump sum worth more than the profits you’d get from investing it?
That depends on the lump sum, expected yearly profit, and interest rates. In finance there’s an idea called “present value [of future cash flows]” which is what you should pay today to get $X/yr forever. It’s something like PV=$X/f(interest rate). High interest rate -> low PV. https://www.investopedia.com/terms/p/presentvalue.asp This is generally worth understanding because it’ll help you make business-aligned decisions about tech debt. So lower interest rates make investments more valuable, higher interest rates make investments less valuable.
Uber made $17 billion revenue in 2021, and took $25 billion in investment until IPO. Imagine by comparison an alternate universe Uber that sold taxi dispatch software. Alt-uber would make $250MM/yr revenue and cost $100MM to build (lump sum). Alt-uber is more capital efficient (ratio of future profits to invested lump sum), but makes less money in absolute terms ($250MM vs $17BB).
So which is better? With a lower interest rate, the PV($17BB/yr) > $25BB and Uber is worth investing in for people who need to make a large absolute amount of money, even at a low ROI. And it’ll crowd out investment in Alt-uber. At a high interest rate, PV < $25BB, and Uber’s a money-losing investment. Alt-uber’s a good investment at almost any interest rate, and wont have to compete with real-Uber for VC dollars in a high interest rate environment.
So in the 2008-2021 era, we saw new companies (“tech companies”) competing with “traditional” incumbents. This was because of a zero-interest rate environment created by the FED in response to the ‘08 crisis. Before that we saw software companies selling software to traditional companies. Now that the FED is moving to nonzero interest rates in response to COVID-driven inflation, we might see a shift back.
Hillel asks engineer-turned-swes whether they think swe is engineering. He should compare that rate to the rate lawyer-turned-swes think swe is similar to law.
Without comparing rates, you might include swe in engineering by an overbroad definition of engineering.
Also if you think engineering is whether the state has granted you a license: 1) cope, 2) since lawyers need a JD, they’re definitely engineers right?
Personally, I think we appropriated the term eng in the mid-90s to up the social status of programmers. Nothing wrong with that, but FAANG SWEs make substantially more than trad engineers now, so we should aim higher.
Kissinger's "World Order" (2014) is much, much better for anyone interested in how the internet will shape the future of governments. 10/10 do recommend.
This book ("The Age of AI", Kissinger, 2021) is as good as a Business Insider article written by a European regulator. Read "World Order" instead! Whatever you think of Kissinger, I guarantee you'll learn something from his 2014 book.
I've been waiting for this forever. Stacked diffs is how I've always worked even before working at FANG, and github just doesn't support them right out of the box.
General Purpose Computing is and ought to be an app on a user friendly internet device. An important app, but one of many.
The fact that the internet device is actually a special-purpose simulator running on general-purpose hardware is an implementation detail. Even most programmers want to do other activities on their devices, like check their email. This should be co-equal with programming; anything you do in your coding environment shouldn't break your ability to get email.
Is general-purpose-computing-as-an-app dying? No: repl.it for kids/consumers is great, there's an explosion of nocode/locode for consumers/businesses; and free tiers of the public clouds are available if you really really want to muck with linux.
Drafting AI offers a different path: teams begin automating processes organically as they discover them, growing toward full automation as understanding deepens. Human review catches edge cases, handles ambiguous situations, and gives an incremental path towards full automation without rework.
Hopefully this lets internal tools teams shift automation further left.