Yes, never seen something quite like it. It's vast, but with enough weird stuff and story sprinkled on it to make it less repetitive.
December 30, 2025 at 10:49 PM
Yes, never seen something quite like it. It's vast, but with enough weird stuff and story sprinkled on it to make it less repetitive.
Just read it. Good stuff! Really liking the art, fits well the story.
Let's see how my understanding of the story old up with time ;)
Let's see how my understanding of the story old up with time ;)
December 26, 2025 at 6:06 PM
Just read it. Good stuff! Really liking the art, fits well the story.
Let's see how my understanding of the story old up with time ;)
Let's see how my understanding of the story old up with time ;)
"grand groupe"
Microsoft, Meta/facebook, Amazon, Apple ... Ou français: Orange, Iliad/Free, Bouygue, Intermarché ...
Choisi une entreprise au pif. 90% de chance qu'elle soit pas sur BS (ou inactive). Avec qq exceptions du dev/jv/recherche, leur public étant assez anglophone/tech pour être sur BS.
Microsoft, Meta/facebook, Amazon, Apple ... Ou français: Orange, Iliad/Free, Bouygue, Intermarché ...
Choisi une entreprise au pif. 90% de chance qu'elle soit pas sur BS (ou inactive). Avec qq exceptions du dev/jv/recherche, leur public étant assez anglophone/tech pour être sur BS.
December 22, 2025 at 5:16 PM
"grand groupe"
Microsoft, Meta/facebook, Amazon, Apple ... Ou français: Orange, Iliad/Free, Bouygue, Intermarché ...
Choisi une entreprise au pif. 90% de chance qu'elle soit pas sur BS (ou inactive). Avec qq exceptions du dev/jv/recherche, leur public étant assez anglophone/tech pour être sur BS.
Microsoft, Meta/facebook, Amazon, Apple ... Ou français: Orange, Iliad/Free, Bouygue, Intermarché ...
Choisi une entreprise au pif. 90% de chance qu'elle soit pas sur BS (ou inactive). Avec qq exceptions du dev/jv/recherche, leur public étant assez anglophone/tech pour être sur BS.
L'ancien twitter. Avec presque exclusivement des comptes personnel (très peu de compte d'institution, pas de grand groupes ...)
December 22, 2025 at 4:06 PM
L'ancien twitter. Avec presque exclusivement des comptes personnel (très peu de compte d'institution, pas de grand groupes ...)
If you enjoy tinkering a bit, I recommend a raspberry pi on your home network + cloudflare tunnel (free tier). Assuming you already have a domain name.
Host whaterver on the RPI, then run cloudflared on the RPI to route the traffic to cloudflare. Safe and easy. And you own it.
Host whaterver on the RPI, then run cloudflared on the RPI to route the traffic to cloudflare. Safe and easy. And you own it.
December 22, 2025 at 8:40 AM
If you enjoy tinkering a bit, I recommend a raspberry pi on your home network + cloudflare tunnel (free tier). Assuming you already have a domain name.
Host whaterver on the RPI, then run cloudflared on the RPI to route the traffic to cloudflare. Safe and easy. And you own it.
Host whaterver on the RPI, then run cloudflared on the RPI to route the traffic to cloudflare. Safe and easy. And you own it.
People using GenAI for ideation are doing it wrong. They should do the opposite of what the AI spit out. The mathematical average of human creation is likely to be boring.
December 17, 2025 at 6:44 PM
People using GenAI for ideation are doing it wrong. They should do the opposite of what the AI spit out. The mathematical average of human creation is likely to be boring.
![A stacked bar chart of Steam playtime by game age. Displaying the following data:
year = ("2022", "2023", "2024", "2025")
# in %
new = [17, 9, 15, 14]
recent = [19, 52, 47, 44]
classic = [64, 38, 37, 40]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:3ambvhier7wbu2png46msydi/bafkreicoisq7wthce74is65em4vttbxwlpzof3c7e2i76tehvqbm6am5nq@jpeg)
Looking at the history:
2022: 17%
2023: 9% (?)
2024: 15%
2025: 14%
2023 looks like an outlier.
2022: 17%
2023: 9% (?)
2024: 15%
2025: 14%
2023 looks like an outlier.
December 16, 2025 at 8:42 PM
Looking at the history:
2022: 17%
2023: 9% (?)
2024: 15%
2025: 14%
2023 looks like an outlier.
2022: 17%
2023: 9% (?)
2024: 15%
2025: 14%
2023 looks like an outlier.
I remember that being a thing for a short while. When buying certain DVD/BluRay you could download a copy. The service was named UltraViolet I think?
November 29, 2025 at 9:09 PM
I remember that being a thing for a short while. When buying certain DVD/BluRay you could download a copy. The service was named UltraViolet I think?
My physical box of Cyberpunk 2077 contains a map of the city, with all the main roads and districts.
And I think most collector edition have a map, at least in their artbook.
But they're not worth using as map, the ingame one being better in every possible way.
And I think most collector edition have a map, at least in their artbook.
But they're not worth using as map, the ingame one being better in every possible way.
November 18, 2025 at 8:52 PM
My physical box of Cyberpunk 2077 contains a map of the city, with all the main roads and districts.
And I think most collector edition have a map, at least in their artbook.
But they're not worth using as map, the ingame one being better in every possible way.
And I think most collector edition have a map, at least in their artbook.
But they're not worth using as map, the ingame one being better in every possible way.
The average steam user play on a entry level gaming laptop (stats are public). Also there's quite a few people that have a Steam Deck, but no (other) gaming PC.
In any case, having a full ecosystem without Microsoft is a good strategic move for Valve.
In any case, having a full ecosystem without Microsoft is a good strategic move for Valve.
November 13, 2025 at 11:38 AM
The average steam user play on a entry level gaming laptop (stats are public). Also there's quite a few people that have a Steam Deck, but no (other) gaming PC.
In any case, having a full ecosystem without Microsoft is a good strategic move for Valve.
In any case, having a full ecosystem without Microsoft is a good strategic move for Valve.
You can use an emulator for non-ARM MacOS. Had to do this once to test a website on Safari. The shit we have to go through to test this godforsaken platform.
You can also rent MacOS servers, a cheap option to build/test. Never used it myself.
You can also rent MacOS servers, a cheap option to build/test. Never used it myself.
October 14, 2025 at 4:05 PM
You can use an emulator for non-ARM MacOS. Had to do this once to test a website on Safari. The shit we have to go through to test this godforsaken platform.
You can also rent MacOS servers, a cheap option to build/test. Never used it myself.
You can also rent MacOS servers, a cheap option to build/test. Never used it myself.
All research show that AI progress linearly at EXPONENTIAL cost.
How can people invest so much on something with exponential cost?
How can people invest so much on something with exponential cost?
October 7, 2025 at 5:24 PM
All research show that AI progress linearly at EXPONENTIAL cost.
How can people invest so much on something with exponential cost?
How can people invest so much on something with exponential cost?
I feel like online, the idea of a "person" shifted to something more intangible, more like a persona.
Also, "person" and "people" are two different words, and my philosophy teacher would say "if the words are different, there must be a reason"
Also, "person" and "people" are two different words, and my philosophy teacher would say "if the words are different, there must be a reason"
October 5, 2025 at 8:08 AM
I feel like online, the idea of a "person" shifted to something more intangible, more like a persona.
Also, "person" and "people" are two different words, and my philosophy teacher would say "if the words are different, there must be a reason"
Also, "person" and "people" are two different words, and my philosophy teacher would say "if the words are different, there must be a reason"
From that, I can guess you finished half the bosses ;)
September 21, 2025 at 10:02 PM
From that, I can guess you finished half the bosses ;)
Such model exists. Look up "infini-gram". But it has rather niche usefulness (Ai2 use it in OlmoTrace).
For general usage, you want a "creative" model. And that require guess, approximation, and randomness. That make it more human-like.
LLM are a lossy compression of human knowledge.
For general usage, you want a "creative" model. And that require guess, approximation, and randomness. That make it more human-like.
LLM are a lossy compression of human knowledge.
September 21, 2025 at 6:56 PM
Such model exists. Look up "infini-gram". But it has rather niche usefulness (Ai2 use it in OlmoTrace).
For general usage, you want a "creative" model. And that require guess, approximation, and randomness. That make it more human-like.
LLM are a lossy compression of human knowledge.
For general usage, you want a "creative" model. And that require guess, approximation, and randomness. That make it more human-like.
LLM are a lossy compression of human knowledge.
This could be a variant of the "Bechdel test".
You can choose a name XD
You can choose a name XD
September 17, 2025 at 5:24 PM
This could be a variant of the "Bechdel test".
You can choose a name XD
You can choose a name XD
If your license truly state "1 device", I would not take the risk. At worst you face revocation.
If you truly need both software running at the same time on two different hardware, your only legal option is having two licenses.
(that's partly why business use remote PC for these)
If you truly need both software running at the same time on two different hardware, your only legal option is having two licenses.
(that's partly why business use remote PC for these)
September 16, 2025 at 4:35 PM
If your license truly state "1 device", I would not take the risk. At worst you face revocation.
If you truly need both software running at the same time on two different hardware, your only legal option is having two licenses.
(that's partly why business use remote PC for these)
If you truly need both software running at the same time on two different hardware, your only legal option is having two licenses.
(that's partly why business use remote PC for these)
(assuming this is about your latest trailer) As a subscriber that didn't watch the trailer: I want finished, full video on my subscription page.
Trailers are to get NEW people interested. So the metric you care about is how many subs this video got you (on YT and Patreon).
Trailers are to get NEW people interested. So the metric you care about is how many subs this video got you (on YT and Patreon).
September 13, 2025 at 2:41 PM
(assuming this is about your latest trailer) As a subscriber that didn't watch the trailer: I want finished, full video on my subscription page.
Trailers are to get NEW people interested. So the metric you care about is how many subs this video got you (on YT and Patreon).
Trailers are to get NEW people interested. So the metric you care about is how many subs this video got you (on YT and Patreon).
I'm not too fond of the tourney in Last Order (or any tourney in mangas, they're way too long), but that vampire backstory is absolute perfection. Definitely worth the read.
I'm also all in for any Nova shenanigans.
I'm also all in for any Nova shenanigans.
September 11, 2025 at 9:13 PM
I'm not too fond of the tourney in Last Order (or any tourney in mangas, they're way too long), but that vampire backstory is absolute perfection. Definitely worth the read.
I'm also all in for any Nova shenanigans.
I'm also all in for any Nova shenanigans.
Difficulty is UX, and as such parts of the accessibility toolbox.
(example: quick successive button press can be impossible for some).
But leaving some people behind can be intentional.
(example: yes, my rhythm game will require quick presses)
Games have the luxury of choosing their users.
(example: quick successive button press can be impossible for some).
But leaving some people behind can be intentional.
(example: yes, my rhythm game will require quick presses)
Games have the luxury of choosing their users.
September 9, 2025 at 4:50 PM
Difficulty is UX, and as such parts of the accessibility toolbox.
(example: quick successive button press can be impossible for some).
But leaving some people behind can be intentional.
(example: yes, my rhythm game will require quick presses)
Games have the luxury of choosing their users.
(example: quick successive button press can be impossible for some).
But leaving some people behind can be intentional.
(example: yes, my rhythm game will require quick presses)
Games have the luxury of choosing their users.
We all know that non-investigative journalism is bad at its job. Yet, somehow, we continue to trust it (Gell-Mann amnesia effect). That's the issue, why do we trust it?
Unsurprisingly, the LLM is better at this particular job when you ask it to summarize a paper, because it fucking read it
Unsurprisingly, the LLM is better at this particular job when you ask it to summarize a paper, because it fucking read it
September 2, 2025 at 9:37 PM
We all know that non-investigative journalism is bad at its job. Yet, somehow, we continue to trust it (Gell-Mann amnesia effect). That's the issue, why do we trust it?
Unsurprisingly, the LLM is better at this particular job when you ask it to summarize a paper, because it fucking read it
Unsurprisingly, the LLM is better at this particular job when you ask it to summarize a paper, because it fucking read it
AGI (or whatever SamA and the like is promising) is the bubble. But GPT is a very real product, with very real users. Interestingly, most of them are happy just using it, and not sharing anything. I'm curious (and worried) on how this will impact media consumption. Similar to the modern content farm
August 17, 2025 at 5:09 PM
AGI (or whatever SamA and the like is promising) is the bubble. But GPT is a very real product, with very real users. Interestingly, most of them are happy just using it, and not sharing anything. I'm curious (and worried) on how this will impact media consumption. Similar to the modern content farm
The problem with that comparison is that NFT/Metaverse never reached broad appeal.
ChatGPT is slowly but surely approaching the billion monthly active user.
Dismissing it as a fad is a very bad strategy to fight it.
ChatGPT is slowly but surely approaching the billion monthly active user.
Dismissing it as a fad is a very bad strategy to fight it.
August 17, 2025 at 4:25 PM
The problem with that comparison is that NFT/Metaverse never reached broad appeal.
ChatGPT is slowly but surely approaching the billion monthly active user.
Dismissing it as a fad is a very bad strategy to fight it.
ChatGPT is slowly but surely approaching the billion monthly active user.
Dismissing it as a fad is a very bad strategy to fight it.