thund

wow, how to romanticize X.400 ...

- poor Internet fit, assuming managed, trusted networks - some promises depended on all participating systems behaving honestly

- once a message reaches another server, you cannot guarantee it isn't copied, backed up, or logged

- X.400 read receipts: more reliable but also more privacy invasive

- X.400 metadata: carries a lot of routing, classification, and organizational info leading to potential privacy leaks

- SMTP is ugly but observable, you don't need a standard specialist to debug issues

If anyone wonders how AI might end up undermining humanity, this is a small preview. We're gradually handing over more processes to automation (YOLO, right?) while the systems themselves remain fundamentally "non empathetic". Maybe this specific case comes down to a poorly designed prompt from 2024... but so far the trend isn't "friendlier" automation, everyone is after more powerful automation...

openai, anthropic, gemini &co should pay wikipedia the rent.

major possible risk, for tech savvy people: another walled garden, unless: - openai will open the memory interface (why would they?) - openai will allow other models (why would they?) - openai will allow a neutral agent/procedural layer (wwt?)

fingers crossed.

Looking forward for the 1Password extension to work properly. Installation works, but the integration doesn't yet. Nice lean UI though, thanks to Chromium oc

4 of them please, going to spend every night playing poker on a real green!

Seems overkill when you can simply tell agents to do that automatically

Plus vscode is maintained by a company with thousands of devs. Obsidian is less than 10 people, which is amazing. About plugins why blame the product, pls check what you install on your machine instead

totally agree, chasing ultrathin designs has diminishing returns. There's so many practical ways to extend battery life such as letting users selectively dial down performance, adding a bit of extra thickness/weight, rethinking how power hungry features are scheduled... It's a matter of priorities, and Apple is obsessed with design (with dubious results, ie magic mouse charging port)

Based on what evidence that adding more battery necessarily makes it a pound? Based on what precedent of mainstream phones ever approaching that weight?

Based on what teardown are you concluding the logic boards take up that much? Based on what phone design history where removing cameras failed to free up meaningful space?

Agree, but by how much on the Pro models? Shouldn’t we expect twice the battery duration every 24 months at least?

It wouldn’t weigh 4lbs if Apple invested on that research. Who really cares about this much thin phones really, Apple is out of ideas, far from revolutionary anymore. They have the hardware to run LLMs, they should invest on batteries and inference quality and consumption with SLMs, rather than delegate AI to Gemini and ChatGPT…

The one I have more vivid memories of is Nokia 8210, pretty small

https://www.businessinsider.com/drug-dealers-are-buying-noki...

https://photogallery.indiatimes.com/gadgets/phones/most-icon...

Yes I forgot about Ericsson, same era.

Funny how these are now classified dumb phones, while they have some odd niche market yet :-)

Around 2000-2005 there was a race for the smallest phone with ludicrously small displays. I believe Nokia was kind of leading and “winning” the race. Then blackberry and iPhone reversed trajectory and suddenly bigger was better, and Nokia died out.

I think we are on the same path here, thinner is not what I want. I want a powerhouse that can run AI for at least 48 hours on the worst conditions, a week at least in an ideal scenario.

This is a result of the current tech being filled with dark design patterns. Tech is designed to be addictive, indispensable, indisputable, mandatory. And at the same time complex, hard, difficult, risky.

We are so used to tech as it is that it is simple to force these bad decisions for the greater good. Because everyone is sure there is no alternative. There’s no other way to design tech, it will always be so complex and powerful that gov and corps can onesidedly decide what is best for the rest of the world.

This might be an area where local AI excels, when ready. No apps. No sharing of personal data. One AI capable of doing what most software does, on the fly, without relying on others to decide what is ok. Remains to be solved who can create and distribute this local AI and whether hardware will be allowed to run “untrusted” AI…

Or US?

fake news, thank you for your attention to this matter! /s

Hate seeing how repos are so polluted by dot-files and dot-folders. I wish tools and SDKs converged into a single dot-folder, so that we could look at repos without seeing everything everywhere

Unlikely but I’d like for this to be true, specially if under the water. Leaving out politics and paranoia for a sec, engineering wise it would be quite an achievement.

It’s time to have self hosted solutions for this kind of activity, with simple UIs, voice editing, text to audio, without subscriptions. Just a matter of time OSS will deliver

Found it, the news reminded me of this paper https://proceedings.neurips.cc/paper/2020/file/747e32ab0fea7...

Is this different than ZipNN? https://arxiv.org/pdf/2411.05239

I see it mentioned but can’t understand if it’s based on it or different/better…

I gave it quick run, didn't work. No biggie anyway, let's wait for local models, they will do whatever we want them to do, stripped of any censorship.

A loved man, he will be missed :-(

Use a function?

    bar(“zoop”)

It’s just syntax, like we used to have

    print “foo” 

that later became

    print(“foo”)