Harmon20 • User • The Register Forums

Isar’s first orbital rocket crashes into sea – CEO calls it a 'great success'

Tuesday 1st April 2025 16:16 GMT Harmon20

Yes, this absolutely can be called "success".

"A trend is starting to be seen in the commercial rocket world where a failure is declared a success."

This is not a trend and it is not in the commercial rocket world only. It's the way it has always been and it is in every engineering discipline.

You build, you test (possibly (probably) to failure) while gathering data, you refine, you build again, you test (possibly (probably) to failure) while gathering data...repeat till the tests reach acceptable levels of performance. If you get good data from the test that advances your design, then the tests were a success. This is how engineering works.

The only difference here is that these companies are doing it out in the open, with the public watching, instead of on closed testing grounds or labs. The fact that many members of said public can't understand that the only measure of a test's success is whether or not it yields good data is neither here nor there.

There is nothing new, disingenuous, or unusual going on here. Suggestions to the contrary just make the suggester look ignorant.

8 0

Guide for the perplexed – Google is no longer the best search engine

Wednesday 18th December 2024 15:06 GMT Harmon20

>>...only when an AI engine gives you the sources of its answers as well as an answer, will Google search face a major threat...

I think that's still a ways off because this isn't part of the core design of the current crop of "AI" engines and the ability to cite sources is beyond their ken. Right now, when I specifically ask the LLMs for sources, they do exactly what they're designed to do - they give me URLs that look very satisfying because they use legit and well known domains and documents so named that they appear to contain the exact information I'm looking for, but they are URLs that point to resources that do not, nor have they ever, existed. (The latter being according to Wayback.)

They're making up their citations just like they're making up the rest of the answers, but that's exactly what they're designed to do. If they were quoting sources that they are aware of and can cite, then they'd be search engines, not AI's. They're fundamentally different tools that do fundamentally different things. They may, and probably will, converge into a single tool doing a single thing eventually, but at that point we won't have two distinct things called "AI" and "search engine" that we're comparing. (I've not tried Perplexity yet. I'll be interested to see if it behaves any differently than I've experienced so far.)

1 0

Meta killing off Instagram, Messenger cross-platform chatting

Wednesday 6th December 2023 20:02 GMT Harmon20

marks matter

The headline punctuation is all wrong. That comma makes it look like it is saying that Meta is killing Instagram, and it is killing off Messenger cross-platform chatting, two things. I clicked into the article to try to understand why I did not hear the deafening roar of the masses at the death of Instagram this morning, only to find out that it was just some mechanical tidbits in the FB ecosystem getting killed off. That comma should have been a hyphen to indicate one thing and not two: "Meta killing off Instagram-Messenger cross-plaform chatting"

1 0