mgraczyk

21 ms·

301.
by mgraczyk 1y ago
Ok so what is the specific annual salary you propose for each police officer? And should they have more influence in local politics than they already do? Currently the police union is either the most or second most powerful entity in San Fr
302.
by mgraczyk 1y ago
But should probably be removed when copying, especially code
303.
by mgraczyk 1y ago
As others mention, these do not look like watermarks, it looks like it's just emitting exotic whitespace in certain cases.
304.
by mgraczyk 1y ago
How do you explain the 9 year delay from 2015 to 2024. And to be clear I am not even claiming that Lurie alone has caused crime to go down. I am claiming that a small number of rich people have, including all of their actions prior to 2025.
305.
by mgraczyk 1y ago
Civic center, but on average it's down a lot throughout the city.
306.
by mgraczyk 1y ago
Then why did crime not go down until less than a year ago?
307.
by mgraczyk 1y ago
No I didn't write that, and I don't believe that's true at the margin in SF. Some cities tax too little and would benefit from taxing more, SF is not one of them
308.
by mgraczyk 1y ago
That's just not true. Our mayor was able to win in part because he is rich, and clearly that has mattered to change the city. Whether that's good or bad is a different question, but it's very obvious a small number of rich pe
309.
by mgraczyk 1y ago
I think we are not disputing the same claim. I claim the city is much better now than 1 year ago, and spending more cannot explain that difference because large increases in spending did not happen 1 year ago and have not helped in the past
310.
by mgraczyk 1y ago
We had 7 years of budgets with that tax before things got better, so it seems unlikely this made the situation better. In fact it seems to have made things much worse.
311.
by mgraczyk 1y ago
Stories like this are fun and I think they resonate well with a lot of people, but unfortunately the details are actively harmful cynicism that ignore fact that would ruin the narrative. Look around San Francisco today. What has changed rec
312.
by mgraczyk 2y ago
How is this not an example of humans being held accountable? What would be the difference here if a help center article contained incorrect information? Would you go after the technical writer instead of the founders or Cursor employees res
313.
by mgraczyk 2y ago
What is the evidence that "dozens of users publicly canceled their subscriptions"? A total of 4 users claimed that they did or would cancel their subscriptions in the comments, and 3/4 of them hedged by saying that they would
314.
by mgraczyk 2y ago
It would also be nice to do that for http for the same reason. You can also read the find Docs for your MCP, and the LLM can also read the docs
315.
by mgraczyk 2y ago
cost isn't part of MCP in the same way that cost isn't part of HTTP. It wouldn't really make sense to include that in the protocol, just put it in the application layer on top.
316.
by mgraczyk 2y ago
My big questions that do not seem to be answered in the paper: 1. Did any participants drop out? Were there exactly 107 treated participants at both the start and end of the trial? Previous UBI experiments I've seen have huge biases wh
317.
by mgraczyk 2y ago
In fact you have it backwards. It has no screenshots at the moment, only the accessibility tree
318.
by mgraczyk 2y ago
It can be, just do that when you install it
319.
by mgraczyk 2y ago
It's a developer tool
320.
by mgraczyk 2y ago
The Reichstag fire has absolutely nothing to do with advertising, and imagining that it does completely ignores and trivializes the entire history of pre-Nazi Germany
321.
by mgraczyk 2y ago
My counter argument to things like this: As you pointed out, very simple registries are already more than sufficient for government oppression. Detailed data that Facebook collects, like which brand of dog food you prefer, is neither necess
322.
by mgraczyk 2y ago
I definitely agree, and I think we should focus on mitigating the actual bad things while either recognizing or considering that the ads themselves are actually good. It's definitely possible to improve the situation and trying to give
323.
by mgraczyk 2y ago
Is there any concrete evidence that anything bad has ever happened as a result of advertising? I don't like rap music, but I think it would obviously stupid of me to claim that it's harmful because I dislike the aesthetics. What i
324.
by mgraczyk 2y ago
Around 2 million people take the SAT every year. A score of 710 on the reading portion gets you into the top 5%. So there are ~100k students who scored 710 or higher on the reading portion of the SAT. Those students are the ones I'm ta
325.
by mgraczyk 2y ago
Based on the intro of the article, the university where this professor works is likely below median. Each year the typical student at his/her university is worse because the best students go to better schools
326.
by mgraczyk 2y ago
> I teach at a regional public university I think this is the main explanation. The median college student has a lower IQ now compared to 10 years ago because more people are going to college, and the marginal new student is below the co
327.
by mgraczyk 2y ago
The original claim was that the new image generation is direct multimodal output, rather than a second model. People provided evidence from the product, including outputs of the model that indicate it is likely using a tool. It's very
328.
by mgraczyk 2y ago
It's not client side, the messages are in the api though. But what do you mean you don't care? The thing you were responding to was literally a claim that it was a tool call rather than direct output
329.
by mgraczyk 2y ago
This is overly cynical. Models typically do know what tools they have access to because the tool descriptions are in the prompt. Asking a model which tools it has is a perfectly reasonable way of learning what is effectively the content of
330.
by mgraczyk 2y ago
The tool name is not relevant. It isn't the actual name, they use an obfuscated name. The fact that the model believes it is a tool is good evidence at first glance that it is a tool, because the tool calls are typically IN THE PROMPT.

More ›