Don’t threaten me with a good time, Mickeysoft!

For those who don’t like paying licence fees, Microsoft is offering a new version of Office, but it comes with some limitations:

Office 2024 packs all the modern features Microsoft has added to its popular apps in the past few years, but that’s about it. The suite comes “locked in” and won’t receive any further upgrades. Users will also miss out on real-time collaboration and Copilot AI integration, which are available with Microsoft 365.

No AI? Oh no! How ever will I cope?

The oncoming enshittification of Discogs

Natalie Weiner writes about the fears that record catalog site Discogs is starting to enshittify itself.

Underlying the sellers’ complaints is a kind of dismay, the feeling that what had previously been a safe haven for nerds to buy and sell $2 records is being threatened — that one more corner of the internet that wasn’t yet a glossy behemoth designed to subsume and capitalize on your personal information was about to collapse.

If you’re serious about music, especially buying vinyl, Discogs is essential. It has the largest catalogue of actually existing records in the world, created over several decades by the users itself. It’s arguably the place to buy obscure records, often cheaper than on Ebay or Amazon. But with it starting to up its fees and other moves, it may be preparing for an IPO or being sold. Even if this does not happen, it’s already in the process of enshittifying its platform, making it worse for users and sellers both.

For me, I mainly use it to get information on music and albums and it’s always my first stop for that sort of information, so it’ll be vexing to see it gone.

Not sure what to think about this

The Washington Post looking into the data sets Google uses to train its chatbots and some surprising results popped up:

A screenshot from the article showing cloggie.org is present in the data set with 230 tokes and rank 11,780,115

As The Post put it:

To look inside this black box, we analyzed Google’s C4 data set, a massive snapshot of the contents of 15 million websites that have been used to instruct some high-profile English-language AIs, called large language models, including Google’s T5 and Facebook’s LLaMA. (OpenAI does not disclose what datasets it uses to train the models backing its popular chatbot, ChatGPT)
The Post worked with researchers at the Allen Institute for AI on this investigation and categorized the websites using data from Similarweb, a web analytics company. About a third of the websites could not be categorized, mostly because they no longer appear on the internet. Those are not shown.
We then ranked the remaining 10 million websites based on how many “tokens” appeared from each in the data set. Tokens are small bits of text used to process disorganized information — typically a word or phrase.

Because the article kindly included a search bar I found out my website is also included in the data set, with some 230 tokens. It would be interesting to see what exactly was included in those 230 tokens from the more than two decades of rambling contained in this site. Sadly, that’s not provided here. Nevertheless, an interesting look in the data used in training socalled AI programmes and its limitations.

Hack into the mooframe

So much for all those visionary cyberpunk authors of the eighties and nineties. They never so much as hinted that in the 21st century, the brave new hacking elite would consist of Nebraska farmers desparate to repair their John Deere tractors:

Not to be too disparaging, it’s hard to predict things, especially the future. But still. Reading that American farmers are buying Ukrainian software to hack their tractors really gives me the feeling of living in one of Bruce Sterling’s novels, which is not necessarily a good thing…

Stupid internet connectivity bug #firstworldproblems

So I’ve been having an annoyingly stupid connectivity bug the last day or so, where when I click on a link my browser starts thinking to itself, then gives me an error message that the site ain’t available. Then, when I try again, up it pops. This happens intermittedly, even on pages I just visited and in both Opera and IE. It doesn’t matter whether I connect to my router through ethernet or WLAN, whether both are used or not and doesn’t seem to happen on my other, laptop computer running Vista (work made me use it, don’t judge me.) My own computer is running Windows 7/64 bits. For one glorious moment I thought it could perhaps be the fake LAN network Oracle VirtualBox needs to have internet access, but disabling that didn’t help either.

Any ideas?

(Windows 7 is decent enough as an operating system when everything works, but once something goes wrong it’s pulling teeth to find out what the fsck is going on. Everything is so locked down and hidden from the user’s view and gets in your way when you’re bug tracking.)