WhiteDawn

Another attraction for publishers was (historically) a much lower adoption of piracy.

Part of the sales pitch is you need more than just a torrent client to pirate.

Now in the world of AI I think this moat is evaporating. Hypervisor exploits are dropping like rain and determined individuals can vibe code a reasonably performant emulator in a couple of weeks.

In that world the only reasonable edge you have left is high prestige exclusives and merchandising/physical sales.

Nintendo has this cornered, one of the few companies who can pull off a rabid fan base buying a console they already own because this time there’s a triforce on it.

Meanwhile Sony and Microsoft are diluting their own IP and going full digital. Writing is on the wall if you ask me.

Looks like they also wrote a Ghidra Decompiler for the ISA https://github.com/DiscoStarslayer/ghidra-spc970

Yeah, 2000 years sounds like a lot till you do the math. Apparently astra was trained on 100k Blackwell gpu’s. So just over 7 days to crack 1028-bit rsa on that cluster…

Not to sound like a mark, I try to not get attached to any of these providers.

I've jumped between copilot, claude, gemini and chatgpt since the start of the year. chatgpt wasn't even worth looking at early this year.

Anthropic has the smarter models for sure, and seems to be default in corporate. However, the amount of budget you get with GPT as a user is much better, the harness feels more polished, and the models are faster. They are also much nicer to work with, I can just read the output for the most part. With claude I get pages of text and need to skim to find where the actual information i need to care about lies. So much more cognitive overhead.

Sol is smart enough for anything I've thrown at it, it's not one-shotting like fable, but I'm more willing to actually go back and forth with it, and it's likely producing better output to keep a human in the loop rather than trying to solve the world independently and making multiple incorrect assumptions.

I think GPT sees the market changing and is correctly repositioning themselves. Anthropic is down the wrong road, and if they don't correct course quickly I'm sure many of those enterprise contracts will start pivoting.

This is all great to get full unfettered access to your own system, as life should be.

I’m sure Xbox and PlayStation security groups are a little nervous right now though. Getting ring-0 on those machines is near impossible, but once you do then everything else becomes wide open

I hope in my lifetime we discover a way to predict earthquakes better, the danger to life from them would be significantly less if there was sufficient notification.

We can alert for storms, tornadoes, volcanoes, hurricanes, typhoons, and even somewhat extraterrestrial disasters like asteroids.

The earth itself must move in predictable patterns similar to the weather, and I'm sure there is decades of research around this I'll never even approach to understanding. Just surprising to have such a big knowledge gap that affects humanity year round

In fairness, games didn’t start employing encryption on the content till the 7th generation. So unless you’re applying ips patches to a 360 game you’re not breaking any DRM

Playing the title you backed up and patched is a different matter entirely

It will be a push and pull and I'm curious where we'll be in a couple of years.

I can fly at 100mph if I let AI run loose and with a bit of steering I can get it to output what I'm looking for and generally pass verification and tests.

If I care about the code though, and I want to keep it maintainable, the amount of time and tokens I need to spend correcting and iterating on the output quickly eats through much of the initial time I saved, to the point where I'm unsure if I'm actually saving much time at the end of the process.

With hobby projects I lean on quality more, and the async nature of AI also makes this much easier to make progress without needing my full attention to do so.

In the corporate world, there's both the pressure to accelerate with AI, but also maintain code and product quality. The dials of one way or the other are more obvious now, but I don't believe it's possible to do both with the current models and harnesses without exponential cost.

What is interesting though, is I'm now leaning towards faster models rather than smarter ones.

Intelligence lets me bite off larger chunks of work at once, and trust the model to behave without having to watch it intensely, but doesn't seem to drive down the number of iterations required to hit my desired quality.

Faster models means the iterations I'll have to go through regardless will complete much faster and gets me closer to a proper flow state. Models will keep improving, but maybe we're getting near "smart enough" and the race will pivot to performance > intelligence.

Clearly hyperscalers are seeing datacenter commitment exceeding demand, and are now running PR campaigns to have the public push back against construction to get them out of their contracts.

Joking conspiracy theories aside, perhaps the over-commitment was part of the plan with the expectation that there will be rejections from government

Thank you for the link to the post. It was a fascinating read and also entertaining.

After seeing the story of the fruit fly brain simulation I figured stuff like the nematode would be trivial.

The lack of active plasticity in LLMs causes a lot of frustration imo. Having agents make the same mistakes over and over and not be able to correct course without prompt engineering is the biggest symptom.

Obviously a static mind is more helpful from a performance perspective, especially if you consider model deployments into gateware.

Maybe this will be an inherent limitation to LLMs that won’t be solvable without a new approach.

Reminds me of this classic: https://xkcd.com/505/

Replace universe simulation with a LLM an it’s hard to consider a bunch of rocks shuffling on a beach as conscious.

Or alternatively, our own minds are simply (squishy, organic) rocks on a beach, which is quite humbling if you think about it long enough.

It’s true, but hosting prices are still within spitting distance of rolling it yourself.

SOTA providers are expecting some level of margin. Companies everywhere have a tight eye on their AI bills right now.

The motivation is there if the models get good enough, even if it’s more painful.

Once someone generates a MTP layer for 26B A4B 4 QAT I'll be singing from the hills with my 5 year old GPU.

Honestly it’s been amazing for me for similar reasons. Also diagnosed ADHD.

Starting projects has always been easy. But once I figured out the hard stuff and then had everything figured out and only saw the long road ahead of drudgery and pipe laying my motivation fizzles out unless my paycheck depends on it.

Now? I still get to figure out the fun hard part and then go send a cheap fast working dumb minion to do the tedium.

I’ve finished 3 things in the past month that have been on my hobby list for years with no progress. It’s been really freeing.

The real moment of truth will be if it’s still worth the cost for tasks that have human value and users but aren’t profitable, which is where most of my side projects live. At current rates it is for me, but once the VC subsidies evaporate then maybe not.

Really their best card is new and additional APIs, building incentives to develop against it.

WinRT (not to be confused with Windows RT, the early ARM version of windows), UWP, GDK, xgameruntime. All of these are relatively new and require virtualization and other security features.

Put pressure on devs by gateing xbox and gamepass behind this runtime and now you have a lever to make the situation more difficult for linux.

Kinda has the opposite effect on me however, as the only reason I'm not subscribed to gamepass right now is the games wont work on my steamdeck. But if MS can get enough killer apps as exclusive to that platform then that will certainly add some pressure.

This is neat, but I think the most interesting part is that it's one of the only models fully unencumbered from copyright. Everything in the training corpus is public domain, guaranteed by the age of it.

First you need to get through the safety net. I’ve had many productive gpt5.4 sessions hit a roadblock of “ethicality” and pollute the context with multiple rounds of trying to convince it to continue

Regular local agent. Seems like as soon as the context fills up (and it only has about 160k of context so that doesn't take much) it starts to fall to pieces. I even tried using opencode as a harness instead and it causes opus 4.7 to lose all memory every time I hit a compaction step.

I wouldn't mind this change that much if opus-4.7 worked properly in copilot cli. It keeps stopping mid-thought or task and forces me to waste more prompts for no observable reason.

Looks like I'm ending my subscription, good (likely too good, no way my account was even remotely within profitable range) access to opus-4.6 was the only reason I used this at all.

Microsoft put AI, Tabs, a login portal, a 'search with bing' action and text formatting on notepad before a 'redo' button to pair with the 'undo' action.

That says everything about the current product priorities that you need to know.

I really dislike the precedent this sets.

A silver lining if this maintainer ends up being in the right is that any proprietary software can easily be reverse engineered and stripped of it's licensing by any hobbyist with enough free time and claude tokens.

Personally, I'd welcome a post-copyright software era

I’ve used AI for some reverse engineering and I’ve noticed the same thing. It’s generally great at breaking obfuscation or understanding raw decompilation.

It’s terrible at confirming prior work, if I label something incorrectly it will use that as if it was gospel.

Having a very clean function with lots of comments and well named functions with a lot of detail that does something completely different will trip it up very easily.

The Stuntman series is a still painful example.

Blame the odd non-IEEE-754 floating point implementation changing physics enough that AI fails most of the missions which softblocks progress quite egregiously

And similar to olive oil. Once you taste the real deal, it's very easy to identify the fake/watered down stuff by taste alone.

In theory the chipset max is 2GB, but the physical motherboard does not have enough address lines traced to support more than 256MB. Adding more memory would require a PCB redesign

Citation needed? The study concluded in Sept 2023 but no results were provided. Are results on these studies usually so delayed? Perhaps they didn’t get the expected result?

So last minute that the AMD engineers found out Microsoft went with Intel at the Xbox launch party [1]

[1] https://kotaku.com/report-xboxs-last-second-intel-switcheroo...

Great post! I find the quote from Bunnie in the blog pretty relevant:

“The JTAG boundary scan approach was rejected on the grounds that the TRST# pin, used to hold the JTAG chain in reset, was tied active in a manner that was difficult to modify without removing the processor.”

Gives me flashbacks to simpler times where disk based systems lacked any real form of DRM because of the assumption that a consumer wouldn't be able to afford to press their own CD-ROMS.

Maybe still not as easy as burning a CD-R, but BGA rework stations have come down in price and utility enough that they are practical for the semi-serious tinkerer. Most modern designs account for this, but I wonder if other techniques, maybe like decaping or some future unknown, will start to open new, simple, vectors of attacks on our hardware today.

I don't really have a point to make here I guess, just that most assumptions made today tend to not quite work out as expected, and that's kinda neat.