dlevine

There are a bunch of comments about how this is the default experience now. It's only the default experience if you buy the ads-free version.

Since I don't want to pay extra to disable the ads, the trick I figured out is to buy the Kindle for Kids. It comes with a cover, no ads, and a 2 year warranty. You can turn off the kids stuff with one switch, and then it's just a normal Kindle.

If you buy it on Black Friday or one of their other sale days, you should be able to get it for the same price or less than the regular edition would be when not on sale.

I really enjoyed Factfulness by Hans Rosling et al.

Potentially a bit dated now, but A New Kind of Science by Wolfram was pretty eye opening.

The use case I'm working with right now is eCommerce. Landing pages need to be served as quickly as possible to score highly on various page speed SEO metrics. I have seen a number of server-side implementations that need to have most of the common landing pages cached to score better than client-side rendered pages.

Many of these page renders require calls out to third-party APIs, which is responsible for some of the slowness. There are definitely optimizations that can be performed (e.g. caching the results of the backend APIs). But it's easier for less experienced developers to just cache the whole page with a reasonable TTL.

Also agreed that you won't be able to cache much (or as much) if the user is logged in. Logged in pages matter less for SEO, so we haven't focused too much on it.

The first time you server a page with a SSR-framework, it will be slow, and subsequently it will be fast when it is served from cache.

What I have seen is that, depending on the number of pages and frequency that each page is accessed, it may be necessary to pre-cache some or all of the server-side rendered pages to get good performance.

In some ways, React on the server is worse than PHP. Server-side rendering of client-side code adds a ton of complexity, and can even introduce performance issues that must be worked around before the app can be launched (e.g. needing to pre-warm caches).

I'm sure there are valid uses for server-side React, but at the end of the day, using server-side languages such as PHP or Rails (or even rendering pages in Node.JS if you like JavaScript) can be much faster and simpler.

I say this having been involved in an app rewrite that used Remix. The project has taken 2 years (and still isn't quite out), and the engineering leaders who made the decision have agreed that we probably could have rendered most of the pages in Rails and used JavaScript for a limited number of things. With that said. I am hopeful that the project will work well when all is said and done, but there were a lot of tradeoffs associated with going to a server-side rendering framework.

I have been screwing around with ESP01s lately (based on the ESP8266, which is the predecessor to the ESP32). It’s amazing how much power you get in a <$5 microcontroller, plus WiFi.

Looks like a good start!

I'm sure you are already working on this, but it would be nice to be able to filter my search by various criteria (e.g. Beds/Baths/Price/Square Footage). Since you guys are an AI company, this could even be freeform (e.g. what are you looking for?)

And also potentially to be able to save multiple searches.

I have watched way too much Steztix Fix on YouTube - it’s kind of addicting to watch people fix things). On that channel,he manages to fix all sorts of things with commodity tools, including a cheap microscope, hot air rework station, and soldering iron. He isn’t doing any BGA soldering, but there are YouTubers who reball BGAs (that looks frustrating).

I paid ~$100 for my Hakko FX888D, and have had had that for almost 10 years. Looking on the Internet, the price hasn't gone up much. Not sure whether this (for $250 including the power supply) is a class above that. The repairability is a definite plus (assuming parts continue to be available for many years), and all the nerdy features are also cool, but not sure how useful they will be in the real world.

I lived through that time (was in high school and college). It was pretty neat to read about the Internet and web browsers in Newsweek, but we didn’t have it at home for a few years. Then it was just a 28.8 dial up connection that dropped randomly. It wasn’t until I got to university that I had a real broadband connection, and even that was slow compared to what I get on my phone today.

Programming (for me) was mostly basic and a bit of C, although I never totally figured out pointers. By the time I got to university, we had early versions of Java and HTML.

That time was fun, but right now is fun in different ways. The tools are much better. Computing is massively cheaper. Laptops weren’t nearly as good as desktops until the early 2000s. Also, you couldn’t do much with AI back then.

I know multiple people who did 15+ years at Google or other big tech cos, and got let go despite having consistently good performance reviews. Those companies figured they could be replaced with younger and much cheaper people (or not replaced at all since the companies overhired during the pandemic).

Fortunately, most of these people managed to find new jobs relatively quickly, albeit with somewhat lower comp packages.

The other outcomes are a redesign and more configurability plus a bunch of new features. It wasn't really an apples to apples comparison. The non-iFrame version was more of a 1.1, where the new thing we are building is a 2.0. Based on some other projects, I do suspect it would have gone faster if we built it on the old stack.

The development team made the choice to go with Remix (well, the tech lead and VP of engineering). No one had used this tech before. We have subsequently talked about whether it would have been better to do the whole thing with Rails + Hotwire. We have been using this approach elsewhere in our stack, and it seems to be a lot conceptually simpler than rendering JS server side and then hydrating it.

We ended up using a reverse proxy since we wanted each product page to have a separate URL. Basically /shop/* would resolve to our React app, which rendered the correct page based on the URL. You could configure this pretty easily using NGINX or Apache, but our customers were pretty technically unsophisticated so it was too much work for them to do it this way.

In the end, we built a Wordpress plugin since it turned out that most of our customers used Wordpress. This plugin acted as the reverse proxy. We went a step beyond this and did some cool stuff to let them use a shortcode to render our eCommerce menu within their existing Wordpress template.

One wrinkle with ditching the iFrame was getting our CSS to not conflict with their CSS. I ended up putting our stuff within a shadow DOM, which was a bit of work but ended up working pretty well.

I work for a company that enables businesses to drop eCommerce into their websites. When I started, this was done via a script that embedded an iFrame. This wasn't great for SEO, and some competitors started popping up with SEO-optimized products.

Since our core technology is a React app, I realized that we could just mount the React app directly on any path at the customer's domain. I won't get into the exact implementation, but it worked, and our customers' product pages started being indexed just fine. We even ranked competitively with the upstarts who used server-side rendering. We had a prototype in a few months, and then a few months after that we had the version that scaled to 100s of customers.

We then decided to build a new version of our product on Remix (SSR framework similar to nextjs). It required us to basically start over from scratch since most of our technologies weren't compatible with Remix. 2 years later, we still aren't quite done. When all is said and done, I'm really curious to see how this new product SEOs compared to the existing one.

After 20+ years in tech, the only time I have ever seen a developer fired for making a mistake was when he leaked credentials because he used one of our repos as the base for an interview project at another company (on a day when he was allegedly working from home). We had a hardcoded key in our package.json to allow it to access a private repo, and he pushed that to a public repo which had an incriminating name like company-x-interview-project-hisname.

We immediately got an email from Github because they scan for such things. We rotated our keys within a few minutes, and he was gone the next day.

Obviously we should have been more careful with our keys, so all would have been forgiven if not for the fact that he leaked keys while looking for another job on company time.

It has been a decade or so since I read The Singularity is Near. From what I remember, Kurzweil said that technologies will increase exponentially once they become linked to the exponential increase in computing power.

Now that Moore's Law has leveled off, I do wonder what will keep the exponential increase going. For example, LLMs are increasing exponentially in size each generation, but that also involves an exponential increase in cost, and will only be sustainable up to a certain point.

The variant with the 8-core GPU is most likely also a binned version of the 10-core GPU variant. It's a lot easier to enable/disable cores than to produce different silicon for minor spec differences.

It may be that Apple's yield of 10-core GPUs is lower than expected, but those GPUs validate just fine with 9 cores enabled. Or maybe Apple had another marketing reason, such as further differentiating the iPad Pro.

I have been curious about this for a while, particularly in relation to the increasing cost of training LLMs.

I was recently talking to a friend who works on the AI team for one of the large tech companies, and I asked him this directly. He said that each generation is ~10x the training cost for ~1.5X improvement in performance (and the rate of improvement is tapering off). The current generation is ~$1 billion, and the next generation will be about $10 billion.

The question is whether anyone can afford to spend $100 billion on the next generation after that. Maybe a couple of the tech giants can afford that, but you do rapidly get to unaffordability for anyone smaller than the government of a rich country.

It will likely be possible to continue optimizing models for a while after that, and there is always the possibility of new technology that creates a discontinuity. I think the big question is whether AI is "good enough" by the time we hit the asymptote, where good enough is somewhat defined by the use case, but roughly corresponds to whether AI can either replace humans or improve human efficiency by an order of magnitude.

I fairly recently used Shadow DOM for a project. We built a widget that was embedded in a web page, and that page could be arbitrarily styled. The outside page was typically a Wordpress theme, which in most cases did all kind of nasty things with CSS.

Even though there is CSS encapsulation, styles can still leak (cascade?) from the parent element into the Shadow DOM. You therefore need to put a style tag on the root that resets all styles.

Even then, there are some things like CSS transformations that will still affect elements within a Shadow DOM.

Another weirdness is that React modals are usually going to break unless you pass them a reference to the Shadow Root. Most popular libraries have been modified to take such a reference.

These are the most expensive job markets and the company is trying to reduce labor costs.

A relative of mine works for a large tech company. Her manager has 2 open reqs on the team. However, he has been told that he can’t hire anyone in California or New York. This is not to say that there are no qualified candidates elsewhere, but there are a ton of qualified candidates in the Bay Area that he can’t hire. He will probably eventually fill the roles, but it will take a lot longer.

There was a good article on this in the New Yorker a while back: https://www.newyorker.com/magazine/2023/06/26/relyvrio-als-f...

It's sad that the drug didn't work (and I had a cousin recently die of ALS), but it's important to let the FDA do their work.

I upgraded my 10 year old PC to Ryzen 5000 a couple of years back, and have never really looked back. Sure, it wasn't that long until 7000 supplanted it, but my PC is still just fine, and will likely be for the next 7 years or so (the biggest wild card is Windows support).

The only thing I really regret is getting a B450 Motherboard that only supports 1 NVMe SSD (and only PCIE Gen 3 SSDs). I would focus on making sure your motherboard has enough RAM and SSD expansion room, and then buy a big enough power supply for whatever GPU you want to run. You might want to figure out whether Zen 5 will have a new chipset, and if so, will the IO be significantly better. Nothing else is going to make a huge difference.

Not mentioned in the article (but present in the original source) is that non-plugin hybrids have 26% fewer problems than ICE cars - I wonder why this is.

Additionally, even if we could remove carbon from the system scale, it's not like everything would magically reset to where it was. We are seeing changes (e.g. extinctions) that aren't really undoable.

With that said, I'm sure that some of the damage can potentially be reversed, and the planet has some capacity to self-heal over long periods of time.

Hopefully reducing emissions and even partial decarbonization will lead to a less bad outcome in the medium to long-term.

Back when I was in college, people would go dumpster diving for the really old analog oscilloscopes with the green tubes. They would hook them up to their stereos in spectrum analyzer mode.

We would use the ones like this to actually do work in lab.

I went to buy lunch today, and the shop was like "we are only taking cash right now." I didn't have any cash, so I went to a different place across the street. That place was only taking cash because their Square machine was down as well. I mentioned to the guy that the place across the street was down. I saw him go to the place next door and ask if their Square machine was down too.

Kind of crazy how one company's outage can bring down most of the places on a block. Everyone just relies on Square for their payments.

Imperial College has a pretty good course on Linear Algebra for Machine Learning on Coursera. It's part of their Mathematics for Machine Learning Specialization. Deep Learning also has a course that appears to be similar.

The specialization as a whole starts simple, and works its way up to PCA.

https://www.coursera.org/specializations/mathematics-for-mac...

SEEKING FREELANCER | US - Remote-only

Technologies: Wordpress/PHP/React

I work for an e-Commerce company that powers a significant percentage of the online cannabis shopping in the US.

We have a Wordpress plugin that makes it easy for retailers to integrate our menu product into their websites. We had a great Wordpress core contributor who built most of it, but he has gotten too busy with other projects. We know something about Wordpress and have made a few enhancements internally, but are looking for someone with more expertise than us since there is a fair amount of technical complexity to our use case (and a million different Wordpress setups that interact with our plugin).

We are looking for an experienced PHP/Wordpress developer who can act as an advisor (including reviewing some of our code), and who is potentially willing to add some more features and do ongoing maintenance/fixes. This would be part-time project-based work.

email me: dana11235@gmail.com

At my first job out of university, we were deploying applications on Solaris servers. I had been using emacs as a text editor, and these servers only had vim installed. I mentioned installing emacs to my manager, and he said that learning vim would be a "feather in my cap." I took him up on that advice, which was sage wisdom.

I spent a lot of my career using various forks of vim as my primary editor, and still use it when I need to edit a file.

Thanks Bram! This world will miss you.