jgalt212

6 ms·

121.
by jgalt212 4mo ago
> Webhooks are the most common way to receive signals from external systems, but processing them safely is not trivial I see webhooks documented all the time, but I have yet to use them in practice, nor have my customers requested them.
122.
by jgalt212 4mo ago
Are you being a troll, or is there a trend afoot to ignore mobile rendering?
123.
by jgalt212 4mo ago
> WordPerfect, by contrast, seemed to require keyboard templates, a manual, a cheat sheet, and a certain amount of divine intervention. I do recall WordPerfect masters being revered, if not more highly compensated, by the average duffer
124.
by jgalt212 4mo ago
Amazing tech, but this wrinkle is so frustrating... > Distance matters. The farther people are from the epicenter, the more likely they are to get an early warning alert, giving them more time to act before the shaking starts. It’s more
125.
Phones Alerted Millions Before Quakes Shook Venezuela(nytimes.com)
7 points by jgalt212 4mo ago|2 comments
126.
by jgalt212 4mo ago
> What software you can run on your computer can already be restricted. Are laws that are inherently unenforceable even laws?
127.
by jgalt212 4mo ago
> I think it is a pity they’ll likely never open source any of this stuf The more they use cloud-hosted LLMs, the more likely it will get leaked into training data.
128.
by jgalt212 4mo ago
> but the default ring is more or less a single, global, object database for the entire bank. Is this really the case? I'm sure there are plenty of transactions that for umpteen different reasons must not be exposed on a global le
129.
by jgalt212 4mo ago
Not true regarding ML, most ML methods support RMSE even if they are non parametric methods.
130.
OpenAI Codex bombards SSDs with needless write operations(theregister.com)
20 points by jgalt212 4mo ago|1 comments
131.
by jgalt212 4mo ago
37 days maybe with Codex? > "On my machine, after about 21 days of uptime, the main SSD has written about 37 TB," wrote developer Rui Fan, a project management committee member of Apache Flink. "Process/file-level che
132.
Life Insurers Aren't Just Investors in Private Credit. They're Major Lenders(wsj.com)
4 points by jgalt212 4mo ago|0 comments
133.
by jgalt212 4mo ago
As a business owner I assure you I am not overly obsessed with labor costs. I obsess on all costs, and all revenues. VC funded and public companies may or may not take a similarly holistic view.
134.
by jgalt212 4mo ago
That's my real fear, that I have massive indirect AI bubble exposure while having zero direct exposure. Similar to how relatively few Americans had liar loans, but we all got burned by the subprime crisis. And then there were those bad
135.
by jgalt212 4mo ago
I'd much rather run a factory where only 10% of my workers may strike than one where 90% may strike.
136.
by jgalt212 4mo ago
But without a monopsony how do these choose your own adventure unions have any real bargaining power?
137.
by jgalt212 4mo ago
Has anyone run a study on how long you can run an agent as root before irreparable damage is done to the VM? A sort of gambler's ruin for the YOLO LLM Age.
138.
by jgalt212 4mo ago
I hope you found writing that response as disatisfying as I found reading it.
139.
by jgalt212 4mo ago
Since 2024, the traffic to our content based marketing pages is down about 85%. What I don't get is how Google has not been terribly impacted as well by the rise of the zero click SERP. Their SERP ad revenue, which is click-based, mus
140.
by jgalt212 4mo ago
How so?
141.
by jgalt212 4mo ago
as in most matters of public controversy, both things are true.
142.
by jgalt212 4mo ago
still cost prohibitive.
143.
by jgalt212 4mo ago
Outside of games and coding generating enough valid examples and counter-examples to harness the power of RL is cost prohibitive.
144.
by jgalt212 4mo ago
so the utility of the KV cache has nothing to do with available RAM?
145.
by jgalt212 4mo ago
> as input tokens that are already in the KV cache are practically free for the provider, not at today's RAM prices.
146.
by jgalt212 4mo ago
So the Ed Zitron audience like the REPL and the hypesters like the compiler?
147.
by jgalt212 4mo ago
If a standard prompt gives you the wrong or spurious answer, the model is not a good model. In 2026, we should not be having to engineer a summarize prompt.
148.
by jgalt212 4mo ago
I don't think reasoning and agentic is an improvement over what came before. It's like going back to batch processing once you've had access to an interactive terminal.
149.
by jgalt212 4mo ago
I concur. His tone greatly undermines the value of the facts he reports. Sometimes / oftentimes his analysis is off the mark, but I have not found him to be reporting falsehoods or inaccuracies.
150.
by jgalt212 4mo ago
So are Hetzer VMs now only 1/2 the price of AWS rather than 1/5?

More ›