gmt2027

If the enzyme is produced by genetically altered microbes like bacteria or yeast, wouldn't that make it practically useful on some scale?

There's a lot of focus on AI taking away jobs. I believe it will constitute an existential threat to a lot of companies. Highly productive solo individuals and very small teams will outcompete large companies by leveraging AI. When capital requirements, transaction and information costs drop dramatically, the optimal size of a firm will change. I believe they will become dramatically smaller.

There's also Alexa. I've seen no noticeable improvement in 5 years. It is difficult to believe that Amazon has any credible AI expertise to sell.

I'd say stop thinking in terms of employment. When you are 100x more productive, you won't need to be a cog in someone else's machine to build products end to end. There'll be less value for the founder types who assemble capital, teams, and middle management to coordinate them. Think of the professions and trades where people run their own practice: doctors, lawyers, accountants, plumbers and electricians. Once sufficiently skilled, they can strike off on their own and deliver value directly to customers. Sometimes with the help of apprentices and trainees. There are a lot of problems to solve today where the cost of software is too high to be practical.

When our tools get so good that our employers don't need us, we won't need them. In a market where the engineers are competing with the manager class whose core skill is politics and bullshit, who would you bet on long term?

In light of the quote, "Civilization advances by extending the number of important operations which we can perform without thinking about them," I agree that LLMs will undoubtedly transform the landscape of work. By automating tedious tasks, they'll enable engineers to become vastly more productive (10-100x), allowing them to focus on strategic and creative aspects of their projects while developing larger, more complex systems.

While job losses are a concern, I think the more significant impact will be on the way companies operate. As firms exist to economize on the cost of coordinating economic activity, the streamlining and reduction of coordination needs brought about by LLMs will challenge the very foundations of many businesses. In this new landscape, individuals and small teams might outcompete larger organizations.

Freelancers and solo entrepreneurs could find themselves better positioned to compete in the market, driving the rise of smaller, agile businesses that can innovate rapidly and cater to niche markets. This shift will also change the skills needed for success in the field.

Overall, it's an exciting time to be part of this industry. Far from being a time to quit, it's an opportunity to adapt, grow, and harness the power of LLMs to reshape the world of work.

My bad. Posted in the wrong thread.

The work of a master craftsman using the most basic hand tools is still leagues ahead of any amateur with advanced tools, automation, fancy CNC, and lasers. Mass production is where the tech would favour that amateur. We are surrounded by cheap mass-produced disposable products and are well aware of their low quality. No gourmet chef should be threatened by the efficiency of McDonald's. Fortunately, our software systems are custom artifacts with zero marginal cost of reproduction. AI will make competent programmers more productive and take expert programmers to new levels.

One transformation will come when a single developer can do the work of a team of ten today. Like solo content creators who no longer need large studios to finance and organise massive production teams and expensive equipment, we'll see more independent developers and small teams creating innovative software with some artistic character. Without the drag of management and upfront financing, the threshold for success will be much lower.

Another will be a market for lemons. Employers will have to find ways to establish competence at building high quality, understandable and maintainable systems. AI can already pass Leetcode-style tests so these are clearly not the right measures.

This is my best optimistic take on things. The rest of the time I'm worried the AI is coming for our jobs.

The work of a master craftsman using the most basic hand tools is still leagues ahead of any amateur with advanced tools, automation, fancy CNC, and lasers. Mass production is where the tech would favour that amateur. We are surrounded by cheap mass-produced disposable products and are well aware of their low quality. No gourmet chef should be threatened by the efficiency of McDonald's. Fortunately, our software systems are custom artifacts with zero marginal cost of reproduction. AI will make competent programmers more productive and take expert programmers to new levels.

One transformation will come when a single developer can do the work of a team of ten today. Like solo content creators who no longer need large studios to finance and organise massive production teams and expensive equipment, we'll see more independent developers and small teams creating innovative software with some artistic character. Without the drag of management and upfront financing, the threshold for success will be much lower.

Another will be a market for lemons. Employers will have to find ways to establish competence at building high quality, understandable and maintainable systems. AI can already pass Leetcode-style tests so these are clearly not the right measures.

This is my best optimistic take on things. The rest of the time I'm worried the AI is coming for our jobs.

So the plane has been flying over valleys at record altitude and we see a couple of mountains ahead that we will easily clear. Despite being absolutely no where near crashing, let's throw a couple of crew overboard so that we can continue to climb to new heights and keep the hype engines of our overvaluation going?

It was news at the time because he was already famous. He had a series of videos on YouTube from his course CS231n at Stanford. He also wrote the popular char-rnn article in 2015 and was a founding member of the research group at OpenAI before joining Tesla.

http://karpathy.github.io/2015/05/21/rnn-effectiveness/

The python packaging situation could be better but poetry is pretty good.

I think the real problem with python packaging is a cross-language one. Show me a language that can elegantly handle building C/C++ dependencies across platforms. This is where things really break down. Python is unusual in this because it relies on C/C++ libraries for performance in a lot of domains.

A good number of native apps are unusable without an internet connection.

We just accept that our fragmented software landscape spans several backend technology stacks, web browsers and two dominant mobile platforms that are in constant flux. ES6, Typescript, Swift and Kotlin didn't even exist a decade ago and their constantly shifting ecosystems are nearly impossible for a single developer to keep up with. Higher barriers to entry work well for the large companies that dominate our industry so they continue to churn out complexity as best practice.

Allowing a single developer to deliver customer-facing applications without the context switch and coordination overhead of a team is a big deal. In the 80s and 90s, this was possible. Today, even deployment has become a specialised role. Not every engineer wants to be forced into a backend or frontend role in an agile team with product owners and ceremonies. There needs to be room for the craftsman in our industry. This is a step towards empowering the individual hacker again.

Liveview Native is not hugely compelling if you're only looking at it from the perspective of an established company or funded startup that can afford a team of more than 3-6 engineers and devops. As a solo dev or bootstrapped startup with limited resources, this is a killer feature for elixir and phoenix.

Integrate segment with your application on the web and mobile and you can enable or disable downstream data delivery to any one of the services at https://segment.com/catalog/ without touching your code. This leaves your engineering team free of requests to integrate your product with new services your marketing and analytics teams want to use.

Could this be addressed by an ensemble of GPT-3s fine tuned with different data?

I've wanted a way to track the full context in which code was written or updated for years now.

This would track and save all resources that were accessed as events. I mean online and offline documentation, specs, pdfs, architecture diagrams, code examples and other parts of the codebase, Jira tickets, bug reports, application screens and output, tests run, failing tests, compiler errors and emails. Everything that the human took in as input before outputting that code. What was Bob looking at when he wrote this line of code? If you had been sitting next to him that week, what would you have seen?

When looking at code, a recommendation engine would show that context, using event data and view frequency to compute relevance.

To do this properly would likely need an isolated developer environment separate from any personal use. Another option would be some kind of ML that would filter out irrelevant events or give the developer a chance to manually edit this data at commit time.

A simplified version of this would track the web resources accessed and allow the developer to manually insert additional links, notes and documents.

This is a tool I would pay for. Call it Retrace.

Max Planck famously said science advances one funeral at a time. J. J. Thomson discovered the electron through his work on cathode rays in 1897. Up to that point the orthodoxy was that atoms were the smallest unit of matter. However there was evidence from the mid 19th century. What Thompson actually did was finally come up with incontrovertible experimental confirmation of a subatomic particle. The electron and other subatomic particles were the gateway to the rapid advance of 20th century physics.

The same pattern repeated itself in the early 20th century. Truly revolutionary ideas may take a generation to gain popular acceptance.

They had vacuum tubes in the 19th century, atomic theory and Maxwell's equations by the 1860s. I find it difficult to believe that they were incapable of isolating the electron.

Maxwell's original paper presented 20 equations. Heaviside's elegant formulation simplified these into the 4 familiar vector calculus equations.

I doubt that this has been the obstacle to a unified field theory.

It was widely covered in the media that unmarried women are the happiest subgroup. However, the reported study may have misinterpreted some data.

The study reported that married women were happier only when their spouse was in the room. When their spouse was absent their response was miserable. The problem is that the original survey defined absence as a spouse no longer living in the household.

https://www.vox.com/future-perfect/2019/6/4/18650969/married...

That sounds a lot like Yahya Abdul Mateen.

There was a talk about medikanren which can make deductive connections from a database of facts extracted from medical research. The speaker used it to find a treatment for his son's rare genetic disorder.

https://www.youtube.com/watch?v=Yval98eOyZc

http://minikanren.org/workshop/2020/minikanren-2020-paper10....

Check for odd or even string length?

  "Above the line" if len(s) % 2 == 0 else "Below the line"

Sounds like My Book of Bible Stories. https://www.jw.org/en/library/books/bible-stories/

Coincidentally, I just picked up a Mazda CX-3 last weekend. You definitely want to have filters for standard things like transmission (auto/manual) and fuel type (petrol/diesel). If these options exist I didn't see them on mobile. None of the sites I used had filters for Mazda's own colours which would have been good.

The popups are intrusive. I have limited patience for that kind of thing when shopping online. When there's money to spend (and car money too), my mental Internet filter is in the strictest no bullshit mode. Popups lower trust, go unread and would have sent me back to Google.

Dick is short for Richard

If you're right about the spin for alignment, we should be looking into deep space in the direction of it's axis. Presumably, this direction is erratic but there may be orientations that dominate over time.

A tumbling probe that doubles as a cosmic pointer to home. Cool.

What I took from Feynman's biography is that he would periodically attempt to rebuild his understanding of all of physics from first principles i.e. basic arithmetic, through algebra to calculus, mechanics etc. Looking for new insights and simplifications, he would apply new tools to reformulate and solve older problems. His seminal work in Quantum Electrodynamics used the Principle of Least Action to great effect. Despite initial scepticism, he came to appreciate the power of LA after seeing that it could be used to elegantly reformulate Classical Mechanics.

Doing this even after becoming a world famous physicist led to the Feynman Lectures in Physics - written as much for his own benefit as for his students. However, I am not sure he used this strategy when learning entirely new concepts the very first time.

It makes sense to do this as practice. To build new theoretical structures, building new constructions of the stuff we already know is a great way to sharpen a toolbox of problem-solving techniques and reveal potential shaky theoretical foundations.

The prize was awarded in 1962, 4 years after she died of ovarian cancer. The Nobel is not awarded posthumously.

Except, Hewish did not initially accept Bell's results. He called it interference, claimed it was man made and not worth further attention. There were meetings to which she was not invited and it was a struggle to get him to accept their significance.

Edit: Also comparing a Doctoral student who "helped build the telescope" to a cleaner is rather unfair. Francis Crick was a PhD student when they worked out the structure of DNA.

Sunscreen protects against the ultraviolet radiation in sunlight which is just outside the range of visible light and distinct from Gamma radiation. Gamma rays are penetrating radiation that would only lose a fraction of the intensity after going through a few centimeters of lead.

If this much of the brain is not completely necessary for its function then there are all kinds of questions raised.

What is the minimum viable human brain? If our brains are indeed highly-redundant scale-free networks, would it be possible to eliminate a precisely targeted 99% of neurons without significant degradation? Can we simulate 1% of a brain on today's supercomputers? How about emulating an existing brain? Do animals have the raw hardware to match or exceed human intelligence? Can we download human 'software' into modified animals instead of machines? Are we already being simulated?

This would keep me up at night too.