Netlify wanted to charge me $55 again (they didn't) for HTMHell. They told me that I'm using up more than 100GB of bandwidth, mostly because of the RSS feed (70GB). The file is 2.7MB uncompressed.
Is that possible? Am I doing something wrong or incredibly right? :D
i beg, once again, tiny js libraries to stop assuming i want to npm-install everything or expose all my visitors to some free cdn service. just give me a directory i can put on a web server, ty
Since 2018, Netflix has raised prices on its most popular plans by around 50%, with the Standard plan increasing from $10.99 to $15.49 per month, and the Premium plan going from $13.99 to $22.99 per month.
At what point does Netflix stop gouging its customers?
How high does the price have to rise for consumers to stop taking it?
And finally - what the ever living f**k has Netflix done to justify raising its prices?
There's a mysterious new, undocumented model in the https://chat.lmsys.org/ arena chat tool called "gpt2-chatbot" - you can access it by selecting "Direct Chat" and then picking it from the big select menu there
It's providing responses that feel significantly more impressive than GPT-4, for both factual-knowledge lookup and logic puzzles. It's possible this is a stealth preview launch of something like GPT-4.5
Okay, if you have 250 GB of disk space available, and a few gigs of RAM, #KomootPhoton is impressive. A query like localhost:2322/api?q=rathausstr+17, with no further qualifiers, bounding box, or whatever, returned after 1048 ms.
That's on my laptop. Searching over all of the addresses on the globe.
@scy depends on your actual use case but geocoding (and routing) on OpenStreetMap data sounds so simple from the outside.
But once you go into details and edge cases the complexity is unbelievable.
Just a few things that come to my mind from my past experiences having worked in a geo company:
How do you geo-code an airport? If you take the airport area's center and stick that into a routing engine, you'll most likely end up behind the runway.
@scy Do all countries have concepts like street names and house addresses?
What about language differences and abbreviations?
Turns out OpenStreetMap mapping schemes can change between countries.
The list just goes on and on and on.
If all you want to do is e.g. have street name auto-complete for a city in Germany then go for it. Otherwise I'd never build this from scratch, it's too complex even whole teams struggle here.
@bagder Just wanted to say thank you for the recording! 🙏
I just went back and watched it, and for someone like me who benefits immensely from the same material presented through different channels (documentation, quick starts, library headers, audio & video) it's a very good overview! ✨
Around the time of this story, I was living through a similar situation in my work life (on a much smaller scope, of course, WordPress.com first, Tumblr later).
Back in 2019, working on WordPress, I started finding myself, almost weekly, arguing against people who wanted to take the product we were working at and made it worse if that mean they could squeeze 0.1% more revenue from it
The 0.1% figure is not even a random number: I remember this speciffic A/B test on WordPress.com that was declared a success and shipped to 100% of the users because it increased the free-to-paid conversion by 0.1%. Soon after it was released, I found out that as a side effect, it increased the churn of free users by 20 something %,so I called for an urgent rollback and removal of the change. So I was promptly explained that we didn't care about free-users churn, because finance had calculated the average long-term value of the free users to be something like $2 per year, and the increase in conversion was bigger than what we could get from them.
Everything became about growth hacking. Everything became thinly-veiled dark patterns. In our private dev slack channels, we joked that since it was impossible to make it smaller or less conspicuous, the next thing the growth team was going to ask us to do was to make the 'free plan' button flee away from the mouse pointer when the user tried to click it. We kept making our product worse, we kept consciously crippling the cheaper versions so we could force people to move to the more expensive options.
Back then I was the lead of one of the two dev divisions working on WordPress.com, so my job was mainly to discuss what we were going to be doing, when and how. And I was getting drained by a constant state of fight against a constant wave of shit they wanted us to build. So much than by the end of 2020, the CEO quietly told me to follow the growth team plans and shut up or step down.
So I requested to move to tumblr, because I thought the pastures were greener over there. But it was all the same: Adding login walls to what we were pretending to be "the last bastion of the free internet", cramping in embarrasingly obvious money-making schemes disguised as features, and making them silently opt-out instead of opt-in so the less people the possible would deactivate them, having to fend off the pressure from the CEO to make everything algorithmic timelines because, you know, tiktok makes a lot of money and why aren't we, etc etc.
I found myself in a place where building something good that people enjoy using was no longer a priority, but tricking people into generating more money for the company was. And when I looked around me, I could see that happening everywhere else, not only in my company. Experiencing the start of the enshittification years from inside wasn't easy.
And, as in the article, the people who decided to turn the shit-metter up to 200%, have a name, in every case. And these people, no matter if they are called Sundar and Prabhakar or Matt and Mark, are destroying the internet. These people are milllionaires, or billionaries, and are destroying our shared, common spaces to squeeze some extra cash from us.
That's why the fediverse and its principles are important. Because that's how we take back internet from their dirty hands. That's how we make internet resilient against them. That's how we build the commons.
@javi agree with everything you say here except that it's the folks at the top to blame.
They're also just doing it for their higher-ups: for the stakeholders, for their investors; and in the end for capitalism. It's a systemic problem, isn't it?
If you're building a CLI tool that can churn on large amounts of data for hours and you don't implement any kind of progress output, we won't become friends.
(And no, it refuses to work with stdin, else I would've just used pv and be done.)