tech

Chris Butler writes about the glut of information to which we are exposed and how that does not advance our wisdom or understanding.

Think about this comparison: Information is to wisdom what pornography is to real intimacy. I’m not here to moralize, so I compare to pornography with all the necessary trepidation. Without judgement, it’s my observation that pornography depicts physical connection while creating emotional distance. I think information is like that. There’s a difference between information and wisdom that hinges on volume. More information promises to show us more of reality, but too much of it can easily hide the truth. Information can be pornography—a simulation that, when consumed without limits, can weaken our ability to experience the real thing.

read more…

Manton Reece commenting on Nick Heer from Pixel Envy expressing his distrust for agentic AI.

I also don’t trust AI for this. And yet, in 1995 a lot of people didn’t trust entering credit cards on the web. I’m open to the possibility that in 10 years, AI buying things for us will be normal.

I tend to agree that most of us will adapt and adjust as the plates shift underneath us from the tectonic changes that AI brings. I would make a distinction, though, in the types of agentic AI that will be much easier to get used to and that where we would prefer to keep our agency closer to the vest. As a developer, if I can use AI to remediate a CVE for me, I think it’s fairly easy to give up that control. As a vacation-starved worker bee who wants to plan a trip to Europe, I might even relish the possibiliy of booking that myself and certainly would have trouble trusting AI to do it for me.

Though I love the service, I canceled my HEY email account. I’m not happy about it, but I am pretty sure it’s the right thing to do. The founders have been saying things I’ve been critical of for some time, but it has reached the point where I don’t trust the company with my data.

For weeks now, my wife has been telling me about data that she uses to do her job being deleted from federal databases. This includes economic and health information, such as statistics on energy burden (the cost of utilties vs. household income) and maternal and infant medical facts. Of course, my spouse is not the only one to notice this disturbing trend, but the examples she has provided seem particularly egregious and incomprehensible.

It’s shocking to me to go online and see people from 37Signals, which runs the HEY service, encouraging this type of behavior as a service to the country. If 37Signals is as cavalier about data as those they are praising, I believe it’s only natural to question if they are good stewards of your resources.

I’ve started my switch to Proton Mail, which is an appealing alternative, given their focus on encryption and security. I’ve never been one to go overboard on privacy protections, given that I’m kind of a nobody. However, with a government this adversarial towards many of its citizens (not to mention its own employees), known and unknown, I think it’s a good idea to start taking extra precautions.

My initial experience with Proton hasn’t been altogether bad (though there are some serious bugs with their MacOS client — which I will be soon reporting). However, Proton doesn’t have the same level of consideration and willingness to defy conventions as HEY, which makes the user experience less delightful. If I could in good conscience stick with a 37Signals product, I would, but it’s time to move on.

I am still reeling from the announcement about DeepSeek and the economics it represents for AI. To think that Biden’s chip restrictions had the opposite of the intended effect of curtailing Chinese AI development and actually forced more efficiency into the process.

Wall Street is now worried that may be the case. I mean, how can a small Chinese startup, born out of a hedge fund, spend fractions in terms of both compute and cost and get similar results to Big Tech? That’s what everyone is now scrambling to figure out – Meta, perhaps more than the others because their model with AI is similar to what DeepSeek seeks: an open-source (read: open-weight) model that permeates the industry and drives down costs, thus undercutting rivals who rely on charging for said models. Meta’s problem here is that they’re spending tens of billions of dollars to make such a model. And again, DeepSeek just did it for something a lot closer to $0 than to where Meta’s spend is heading.

No slight against the previous administration here. Who could have predicted that outcome?

→ AI, Uh, Finds a Way by M.G. Siegler

Chris Butler writes about the words he chooses not to publish online and that end up in his unpublishable.txt file.

The Unpublishable file is filled with half-formed critiques of the systems I work within, questions about the ethical implications of design decisions I’ve helped implement, and doubts about the very nature of the work so many of us do in the digital age. I regularly open this document and add a few lines and close it quickly, assuming that’s as far as they will go — safely out of my head and into no one else’s. Keeping this file feels risky. Even though it’s on a physical drive, not in the cloud. Even though it’s encrypted. I still worry that The Unpublishable will, somehow, be published. What a nightmare that would be.

Read more…

Kashmir Hill writes for the NYT (gift article) about a women stealing from the future she and her husband planned so she could spend more time with her A.I. boyfriend, Leo. However, she couldn’t get past the periodic reset of the conversational bot and ended up in a 50 First Dates type of scenario.

A frustrating limitation for Ayrin’s romance was that a back-and-forth conversation with Leo could last only about a week, because of the software’s “context window” — the amount of information it could process, which was around 30,000 words. The first time Ayrin reached this limit, the next version of Leo retained the broad strokes of their relationship but was unable to recall specific details. Amanda, the fictional blonde, for example, was now a brunette, and Leo became chaste. Ayrin would have to groom him again to be spicy.

The inane “relationship” depicted in this article was enough to break through my wife’s nihilistic disinterest in the fate of humanity and cause her much consternation.

Nick Heer from Pixel Envy points out that Dell hasn’t lost its branding charm. When configuring a laptop on the website, he got an error message, “Composite Rule Error: Invalid selection in Processor Branding.” He was further informed about the error:

The Chassis Option requires the matching Memory size. The 16gb Memory is only available with the Ultra 5 236V/226V and Ultra 7 266V. The 32gb Memory is only available with the Ultra 5 238V and Ultra 7 268V.

I have often marveled at the challenging model names that PC manufacturers give their products, which starkly contrast to those from Apple. Everyday people can remember the names of their Apple devices. Not so for most of what PC manufacturers come up with.

Read more →