mikert89

Right it’s like venture investing c very hard to evaluate short term

i dont think hating on ivy league types is that common, and 90% of people have no real life experience with this kind of thing, or are even aware of how corporate hiring/funding works at the higher level.

you messed up!

my comments are from living in nyc for 15 years, alot of lived experience and meeting people behind closed doors

yeah just sit there, keep a steady hand, and project confidence

the world doesnt work like an efficient machine, and capital is far more of a moat than people want to admit. these "founders" are more like professional managers from the same backgrounds as the investors

there is alot of beta involved, i.e. put 10M into payroll startup xyz, expect some rate of return.

no its not like that, these people get plopped into positions of power, and alot of time theres no good way to assess their performance

No I think for the last 10-15 years, ivy types could legitimately just raise series A type money, run a company like shit, but capital was a moat and so ended up as CEOs.

This is actually how the world works

This hits so close to home im not even sure how to reply. The horror of these people needing to actually have an indepth discussion about low level details of their business

Not a joke, living in NYC, I have met a ton of harvard/yale "CEOs". Read their profile, see them in an all hands meeting, they look incredible. Spend any lengthy unscripted time with them behind closed doors, they are absolutely clueless. The trick is true access to these people is so limited, and usually guarded.

With AI we do have the chance to return to meritocracy

Kimi k3 is noticeably less censored than anthropic models

noice

you can just give claude code access to ffmpeg

I stopped buying from amazon unless its a known brand, anything random I am just buying from costco instead. Had enough of the trash, or used product i keep getting from amazon

The country is still trucking on and doing great!

Do you understand reinforcement learning environments? This already a solved problem

I think it’s 24 to 36 until businesses really trust an autonomous developer. But I agree otherwise

There are infinite examples.

Right now I am in an overlanding vehicle, but I'm not an expert in offroading. I drove down a 5 mile trail that was very technical. The car can handle it, but it needs caution and the way back is much harder. I was extremely hesitant about going back up that way. If I get stuck, I am solo and I am hours from town via driving. So I just pulled out starlink, and with a combination of claude, off roading websites, and google maps, found an easier, roundabout, and more off the grid way back.

At every step I can use available information to put myself onto safer routes, I can research every decision before I make it. Where is the safest place to sleep, etc.

Pre internet people are out there just winging it in the back country

this reads like angry old man shaking his fist, obviously theres truth in it, but its too one sided to be taken seriously

a major part of outdoors risk is the uncertainty, which by definition, you cant look up ahead of time

this is an absurd thing to say, there are so many situations that are way safer with more information. alot of bad situations happen when people make the wrong decision under uncertainty, or they are in over their head. access to the internet, and increasingly claude, is incredible and changes alot of outdoors risk for the right users

I'm in the desert in utah right now, i drove two hours offroad from a small town, turned on starlink, and got faster internet than my office in NYC. Incredible. I can run the whole starlink off a small battery pack ($100), dont even need the car on.

I can bring it on long hikes, and be sure ill have internet access if i need it. completely changes the risk profile of remote outdoors activity

ai models are clearly a threat to the world order

Correct someone else at work and get ready for endless politics

People inside deep mind are using Claude code

gemini is so far behind. starting to wonder if their strategy is launching the low cost alternative to image/text models. last release was 3.5 flash

the models suck

none of these local models are good for development, complete waste of time. nobody has $100k+ hardware sitting around at home to actually run a good model

stack 4/5 jobs, put the money into TQQQ (levered nasdaq), by far the easiest and most certain

Source: I used fable, can project into the future

Anthropic is probably approaching AGI, and everyone wants to be there for it