peterbonney

6 ms·

31.
by peterbonney 3y ago
> I can't trust the answers it provides or the text it generates. It's not a replacement for search, it simply makes search worse. Can you trust search results without AI summarization? They’re mostly SEO spam. > The images
32.
Telling AI model to “take a deep breath” causes math scores to soar in study(arstechnica.com)
8 points by peterbonney 3y ago|4 comments
33.
by peterbonney 3y ago
Fentanyl is frequently laced into other drugs (without the buyer knowing) because it is cheaper to produce than e.g. heroin. Or it is used as the main ingredient in fake versions of “real” painkillers like Oxy. But it’s so deadly at low lev
34.
by peterbonney 3y ago
If “allowing the execution of arbitrary database queries written by an LLM inside a SaaS application” is the answer, I’d love to know what the question is.
35.
by peterbonney 3y ago
Reminds me of this incredible New Yorker article about illegal mining in abandoned South African gold mines: https://www.newyorker.com/magazine/2023/02/27/the-dystopian-...
36.
by peterbonney 4y ago
That is one side of a conversation where the other party as a rule never comments. Coinbase is simply not a reliable source for characterizing the SEC’s position on this.
37.
by peterbonney 4y ago
At the risk of stating the obvious, she didn’t actually say anything that implies more than “the FDIC is doing its job” (helping depositors). There’s a lot being read into it that I don’t think is there.
38.
by peterbonney 4y ago
Yeah, it’s definitely a major hole in the accounting. Having said that, heat in Manhattan tends to be very efficient because of the density. The majority of the radiators in my apartment (it’s a prewar building with old-school cast iron rad
39.
by peterbonney 4y ago
I have one thing to say about this: “Online survey”
40.
by peterbonney 4y ago
What you say about most managers is true. And yet, still the situation I described can arise for perfectly valid and rational reasons. Like if the context required for your 30 seconds of input is difficult to convey quickly - if you’re avai
41.
by peterbonney 4y ago
Sort of an obvious point, but the opportunity cost analysis here has basically nothing to do with whether the meeting attendees are developers - it applies to anyone who has other things they could be doing with their time and has a cost fr
42.
by peterbonney 4y ago
It’s wild that Google, despite being the leader in startup email/business software and literally giving their Slack competitor to everyone for free as part of their Google Business subscription, STILL has such low mindshare in this spa
43.
by peterbonney 4y ago
> didn't realise that they can get personally liable for billions of dollars of loss for shareholders and go to court for years if they don't at least consider the offer It is ludicrous to believe that, even if this were true B
44.
by peterbonney 5y ago
My broken-record reaction to headline-worthy research findings: The past decade of social science research publication has proven that provocative-sounding results should be considered fraudulent until the underlying data has been published
45.
by peterbonney 5y ago
I know someone who makes her own dyes to color her own yarn for knitting. And yet with vanishingly few exceptions, not even the most hardcore crafters want to dye their own yarn. Thus it is for servers.
46.
by peterbonney 5y ago
The best you can come up with for controversies in epidemiology are from non-epidemiologists?
47.
by peterbonney 5y ago
Does epidemiology have a well-publicized replication crisis with numerous examples of outright fraud from high-profile researchers? If so then yes. If not then why don’t you just say what you’re trying to say rather than imply it obliquely
48.
by peterbonney 5y ago
I think I say this every time I see a headline like this on HN, but I'll say it again: The past decade of social science research publication has proven that provocative-sounding results should be considered fraudulent until the underl
49.
by peterbonney 5y ago
> …all of the SEC’s questions in writing and then again in person… They asked for documents and written responses… Plain as day! And since you have so much experience dealing with the SEC, you surely understand that the specific question
50.
by peterbonney 5y ago
They repeatedly refer to lengthy sets of questions they received from the SEC, both in person and in writing. If you don’t think that is the SEC communicating their “rationale” for a decision and offering insight into what acceptable altern
51.
by peterbonney 5y ago
It is clear from their own blog post that they have communicated with the SEC in considerable detail, and the SEC has told them in no uncertain terms not to proceed. It sounds to me like the SEC welcomed the discussion, it’s just that Coi
52.
by peterbonney 5y ago
“We have no idea why the SEC intends to sue us! It is a complete mystery.” (Later in the post…) “All we’ve done is create a thing that people can trade which we are moving towards selling to investors against specific SEC advice to the cont
53.
by peterbonney 5y ago
Scientists can’t develop a null hypothesis about the impact of the most studied virus in the history of humanity? If you have no null hypothesis then you shouldn’t be drawing conclusions from data, period. Anyone who reads this site regular
54.
by peterbonney 5y ago
I couldn’t have guessed that we could find two samples of people such that antibodies in one group exceeded antibodies in the other? That is true for any non-uniform data set. As for your other point, I also maintain that the opposite findi
55.
by peterbonney 5y ago
What does it add to our knowledge? That some sample of people have more antibodies than a non-equivalent sample? I could have told you that without a scientific study. That same sample of people who had the illness may have had more, fewer
56.
by peterbonney 5y ago
You’re missing my point. We know nothing about the sample that was excluded (the people who died). Were the survivors already more prone to wearing seat belts, or did the accident change their behavior? In the Covid case, is the sample more
57.
by peterbonney 5y ago
It’s math for the sake of math with no actual scientific value. Here, let me help: “Car crash survivors show greater seat belt compliance than people who watch a seat belt PSA.” Useless analysis.
58.
by peterbonney 5y ago
The term ”survivor bias” is usually used metaphorically, but in this case it applies literally. Putting in some caveats about the risks associated with infection is not enough. This is a garbage analysis that should never have been accepted
59.
by peterbonney 5y ago
I don’t - I was just speaking abstractly. In practice I can’t recall ever meeting a dev with elixir experience who seemed unqualified. And I’m using “seemed” rather than something more definitive because I’m referring to a broader set than
60.
by peterbonney 5y ago
If you’re on the job hunt and you’re interested in Elixir, I can guarantee we won’t reject you through an automated resume screen (I think they’re terrible and we don’t use them :). My email is in my HN profile - feel free to reach out!

More ›