reliefcrew

Joined 4 years ago

Karma
133
Submissions
196
Member since
Mar 11
in base64…

cmVsaWVmY3Jldy5uZXQK

View on Hacker News

Recent submissions

Comment · 7 months ago
> There was no clear pop-up, no "Are you sure you want to

You haven't learned MSFT's m.o. yet... this kind of pop-up only shows up when you're absolutely certain about something, like removing files. In those cases they ask for confirmation endlessly and needlessly.

When they're trying to collect money, they go ahead and quickly guess with all the defaults as they see fit. I'm surprised the default isn't 1000 seats tbh. After all, you're certain to be a huge success now that you've chosen them as a vendor.

;-)

Comment · 8 months ago
Yes, they have data many don't have access to. And, they make "educated guesses" based on this data as well as the expertise available to them.

Their predictions may certainly be embellished as well. However, their statements can also be taken out of context or misinterpreted.

None of this is in question. The challenge is to make a careful judgment of your own.

Comment · 8 months ago
Some things never change :-)
Comment · 8 months ago
> Say hypothetically the productivity boost from AI is 2x per employee

I believe this is a bad assumption. Most likely the boost is not normally distributed. So, although all employees become more productive, some employees become much more productive thereby making the others redundant.

As someone pointed out already... there are diminishing returns. In addition, there are marginal returns. Put the two together and you get ever diminishing marginal returns from the laggard employees.

Comment · 8 months ago
A carefully audited risk management system should always sit in between trade generation and execution. So, direct broker access is definitely a terrible idea.

Vibe coding the generation of simulated trades is another matter. That has value.

Comment · 8 months ago
https://mitpress.mit.edu/9780262560993/the-little-schemer/

She can play along on the computer by installing mit-scheme (http://www.gnu.org/software/mit-scheme/) and placing the snippet below into ~/.scheme.init

   (define atom?
     (lambda (x)
       (and (not (pair? x)) (not (null? x)))))

   (define lat?
     (lambda (l)
       (cond
        ((null? l) #t)
        ((atom? (car l)) (lat? (cdr l)))
        (else #f))))

   (define member?
     (lambda (x l)
       (cond
        ((null? l) #f)
        ((eq? (car l) x) #t)
        (else (member? x (cdr l))))))

   (define mber?
     (lambda (a lat)
       (cond
        ((null? lat) #f)
        (else (or (eq? (car lat) a)
                  (mber? a (cdr lat)))))))
Comment · 8 months ago
Oh, yes... I didn't mean to imply that the newer "AI modes" aren't better and that we're somehow standing still. More that the differentiation is made much starker due to so-called "enshittification". Whether that contrast is deliberate or coincidental I'm not completely sure, but I'd wager it's the former.
Comment · 8 months ago
It seems the relentless pursuit of greater profit has diluted "the old search interface" toward simply regurgitating adverts.

Customers realize this, even if subconsciously, and so the new "AI Modes" offered by search companies are required to stay relevant.

It's the process of creative destruction at work... round and round we go.

Comment · 8 months ago
> And if they fix it? Pay them. Simple.

> Stop fooling around with abstract puzzles

I beg to differ on the simplicity here.

The very best candidates don't want to think about your stuff at all, unless they're paid ahead of time, since they've already made tons of $$$. In effect, they're already getting paid by your competition or they themselves are your competition.

It's a great example of Akerlof's "market for lemons".

The "satisficing" solution IMHO is to take [calculated] risk on available consultants/contractors and keep the better ones on retainer [to hire]. Yes, in a sense, I'm suggesting you pay [more] to interview people. That way you stand a chance of beating the average.

Finding the "best value" in a candidate is a trap... you only need good enough.

Comment · 8 months ago
If the sender eventually proves to be a freeloader, they get added to an auto-respond list. I hope to enhance the auto responder with AI soon but for now it just tells them I'm busy and to bugger off. You never miss anything that important.
Comment · 9 months ago
> Is there anything I can do?

Have you tried reading Camus' essay "The Myth of Sisyphus"?

Comment · 9 months ago
It certainly depends on the kind of writing too. Very technical writing should benefit from a computer. Poetry... probably not as much.
Comment · 9 months ago
Obviously I can't speak for op, but... spellcheck and copy/paste alone make paper more difficult to work with.

Basically, writing involves a lot of re-writing and revision. Electronic documents make that easier.

Comment · 9 months ago
> The main practical win is that cmd args show up in `ps aux` for anyone on the system to see, whereas stdin keeps it off that list.

For those interested, re-mounting /proc with hidepid can prevent this:

    `mount -o remount,rw,hidepid=2 /proc`
Comment · 9 months ago
> I don't know how sleep/wake works from a text mode TTY machine if that's what you mean.

Usually, I expect, by just issuing a command... e.g. `systemctl suspend`

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton