Hacker Newsnew | past | comments | ask | show | jobs | submit | SketchySeaBeast's commentslogin

Is Opsec 101 to increase the estimate by two orders of magnitude? "We think this operation will take about 10 weeks, so we're estimating 10 years."

The answer is "long enough to avoid giving away operational details," not some robotically applied constant multiplier like 10x.

We also don't know whether they expected this to take 1 day or more. Just because it worked out quickly doesn't mean that's the "worst case" operational timeline.


Isn’t that how estimating timelines should work?

I think it would be more accurate to say "based upon true events".

Wait a minute...

3. They discovered it's something they can measure so they made a metric about it.

4. They heard from their golf buddy who heard from his racquetball buddy that this other CTO at this other shop is saving lots of money with AI

I know you're speaking half in jest but the C-suite of my area actually used a tweet by an OpenAI executive as the agenda for an AI brainstorm meeting.

Well that's inspiring. If you're going to follow anyone right now be sure to follow someone from the company that has committed to spending a trillion dollars without ever having a profitable product. Those are the folks who know what good business is!

It'll never cease to amaze me how many powerful people can't tell advice from advertising.

I am at less than half jest here.

I have friends who are finance industry CTOs, and they have described it to me in realtime as CEO FOMO they need to manage ..

Remember tech is sort of an odd duck in how open people are about things and the amount of cross pollination. Many industries are far more secretive and so whatever people are hearing about competitors AI usage is 4th hand hearsay telephone game.

edit: noteworthy someone sent yet another firmwide email about AI today which was just linking to some twitter thread by a VC AI booster thinkbro


This feels like a construction company demanding that everyone, from drywaller to admin assistant, go out and buy a drill.

Can I modify your example to:

Demanding everyone, from drywaller to admin assistant go out and buy a purple colored drill, never use any other colored drill, and use their purple drill for at least fifty minutes a day (to be confirmed by measuring battery charge).


Better, yeah.

Awesome, with that new policy we'll be sure to justify my purple drill evangelist role by showing that our average employee is dependent on purple drills for at least 1/8th of their workload. Who knew that our employees would so quickly embrace the new technology. Now the board can't cut me!

It's really cascaded down too.

Each department head needs to incorporate into their annual business plan how they are going to use a drill as part of their job in accounting/administration/mailroom.

Throughout the year, must coordinate training & enforce attendance for the people in their department with drill training mandated by the Head of Drilling.

And then they must comply with and meet drilling utilization metrics in order to meet their annual goals.

Drilling cannot be fail, it can only be failed.


This is literally happening in non-tech finance firms where people in non-tech roles are being judged on their AI adoption.

Some companies swear by this. CP Rail is notorious for training everyone to drive a train.

That kind of makes sense philosophically if your business is trains, but I don't think that their business was AI agents. Although given they have a VP of AI, I have no idea. What a crazy title.

I'm not a huge fan of conspiracy theories, but starting a 240 hours closure, ending it after 4, and claiming it was a test? What sort of testing are they doing that they were off by two orders of magnitude about the duration?

Who knows. Maybe the system was malfunctioning and they didn't know how long it would take to shut it down. More likely the admin is just blabbing the first thing they think will shut everyone up.

Someone probably briefly thought they brought Skynet online via AI powered drones.

The medevac is scary. This could cost lives. Hopefully there's a reason but, given how clumsy this admin is, who can say?

I guess it depends on your definition of "fluid" intelligence, though I was bad with both of them in my 20s.

Those of us that grew up stupid have the advantage here - our coping mechanisms never stop working! Everyone else has to relearn how to make it work.

Both of you take your well-deserved upvotes and scram.

But that's not following the saying - it's still not proving, it's modifying the rule. It shifts the rule from "birds can fly" to "most birds can fly". Pointing out that penguins can't fly doesn't make the case that birds can fly stronger in any way.

You're right in a strict sense. But in my experience such strictness is only useful in hard sciences and (maybe) legalese. There are exceedingly few things we can claim to apply everywhere, and even fewer we can "prove" to each other.

Give it a try if you don't believe me. Even categories we take for granted, like trees and fish, are not perfectly crisp, and "obvious" facts like "humans need a heart to live" have surprising exceptions.

> Pointing out that penguins can't fly doesn't make the case that birds can fly stronger in any way.

I disagree. It's such a common rule that there's a long Wikipedia page for the exceptions[1], and the first photo is of penguins, labelled "penguins are a well-known example of flightless birds.".

If I knew nothing else about the topic, I would take it as evidence that it's common for birds to fly, otherwise that fact would have been unremarkable. Not hard proof of a universal quantifier, but a useful rule nonetheless.

[1] https://en.wikipedia.org/wiki/Flightless_bird


> There are exceedingly few things we can claim to apply everywhere, and even fewer we can "prove" to each other.

Yes, this is why hard and fast rules don't make sense, and why they should have "generally", "normally", or "mostly" attached to them.

If you have two categories of birds, one with those that fly and one that doesn't, having that second list doesn't make the first stronger. At some point that second list dilutes that first one so much that it doesn't make sense anymore.

If my rule is that "white guys are named Dave" does my building a list of every example of a Dave and non-Dave make my rule stronger? When does the "strong" nature of the rule get watered down sufficiently? Honestly, a list of hundreds of birds tells me that it's a weak rule and that the "birds fly" rule is wrong.


> And, if so, is Gates likely to be investigated by the police?

What a bizarre turn of events that would be if THIS was the thing that got investigated.


It would be a bit like Al Capone and justice by unusual legal means.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: