The web’s ad markets bake in a loss of privacy by design, and so every website that wants to participate in those ad markets has to dupe as many users as possible into handing away their privacy. Couple that with declining web ad revenues because of the move to mobile apps, and you get a web filled with malicious compliance cookie banners.
Had google offered a privacy-preserving ad model (basing the ads off the content of the page instead of who the user is) things would be very different.
Why would they? Unless they have lots of unused tpu real estate that they could host it on “for free” they would be bumping more profitable workloads off of machines to give away that capacity to people with zero long term loyalty. There is no business reason for google to subsidize these models.
OpenAI has too much money. They’re spending their money in stupid ways.
I think they got some data center deals very cheap when no on was thinking about data centers. Dario didnt want to take that risk so Anthropic didnt make the deals earlier but now paying Google and Elon higher rates.
Having people use the model generates real world training data which could be useful. Other than that I think you're right that there is little reason to artificially boost users with subsidized pricing.
For running it in yolo mode I set it up in a devcontainer. It takes a bit of figuring out, but once set up the permission prompts go away and claude has no access to anything I haven’t explicitly added to the devcontainer (unless it hacks its way out).
I tried to figure out devcontainers (I don't use vscode) but it seemed like a lot of complexity and ended up just doing the old-school thing and creating a separate user/group that I ssh into with my main account as a member of that user's group so that I can browse and edit/add files.
This runs the specified agent in a read-only container with only /workspace and /tmp writable. Obviously, you need to prepare the image first from a Containerfile/Dockerfile, with the required toolchain and agent installed. I use agent-specific ssh keys that are baked into the image, but you could also bind-mount specific files from your own homedir if required.
There’s also the Linked Data Event Streams (LDES) standard, which is a way of hosting a log as a set of linked http documents, fetched via polling and link navigation. Is there really anything more needed than a webhook for the notification and an LDES for the log?
You can create anything and everything. Web components are about extending html to make it do more, progressive web components are about rendering that on the server and progressively enhancing with interactivity in the browser. Elena solves the major pain points in doing that. Components built this way can be used in any framework.
Basically what we're looking at by the M7 generation is a tier shift, where the base M7 can do what the M2 pro did, and every tier moves up accordingly, with the M7 ultra becoming competitive with nvidia dedicated consumer hardware.
New cheap android phones are just as slow as old cheap android phones. The bottom of the market has been stuck in performance limbo for years, and modern web dev frameworks are ill designed to meet them where they are at.
Mobile devices rarely last beyond the 5 year mark, although the age span has been increasing in the past several years and the resale market is booming thanks to inflation, geopolitical instability, and a global cost-of-living crisis.
Oh, I know it's not the point but I find it a bit disingenuous going from iPhone base model to the Pro in the last 3 years and still comparing to the base model Samsung S series. Though maybe I'm missing something non obvious.
But yeah, generally I've seen a better experience buying used phones (in good condition) instead of budget/cheap new ones.
The top two lines are the fastest devices available in the iOS and Android ecosystems each year, and when using Samsung S-series devices, I have made sure to pick the faster of the Qualcomm vs. Samsung Semi parts (both are used historically, but Qualcomm is most prevalent in the US). As explained in the piece, the bottom two lines are taken from representative devices in the mid-range and low-end price categories. There's wild variation in those market segments due to short-run discounting, and the most meaningful uplift in CPU speeds has coincided with 5G radios (for obvious signal-processing reasons). Because the lowest-priced segment doesn't yet feature 5G, those processors have remained stubbornly slow, even if they get cheaper every year.
Thank you for clarifying and sharing the full blog post! It was actually an error on my part because until now I didn't realize that the S-series base model and the Ultra model actually share the same CPU. I should've checked my assumptions first.
I also saw you addressed in your blog post the growing market for refurbished phones and in the last year or so I've noticed a similar thing in Romania where a e-commerce giant is involved with a company that refurbishes and resells old user devices to try and corner a big piece of the pre-owned (second-hand) market where previously you'd be stuck hunting for good deals on the national Craigslist equivalent or searching for small-ish repair shops that sold refurbished devices.
Internal to what? The brain is not a monolithic thing, it is different parts communicating and interconnecting. When the connection between the halves is cut, the person objectively becomes two people, but still experiences and presents themselves as one. Observing ourselves is just one part of the brain responding to another, or theorizing on past behavior. There is probably no actual introspection going on inside of the human brain, only the perception of one.
Internal as in that which is feeling things. In this context the physical brain is what is responsible for what you do and feel, but it is not physically different from other matter and it requires no "internal observer" do do its thing.
reply