dostres

I imagine you're making the Engineers mistake of pitching your credentials. Think of the hiring from the their perspective. A 25 yo recent Stanford Grad with no idea what they're doing with crippling insecurity. If they actually hired someone with experience then they might be outed as 'not the smartest guy everyone has ever met' and that would be devastating to their self image.

In order to send a signal that you're a safe hire that won't undermine them, either intentionally or accidentally, you'll need to show that you're a team player who will go along with their ideas no matter how obviously stupid they are. Crossfit is perfect for this.

The US has tax on all income regardless of residency.

I do think it could be possible to use an EEG to get ahead of a mouse click faster than the hands and perhaps cut off some milliseconds in delay.

But I don’t believe that Elon Musk’s Neuralink will ever offer anything of value to a non-disabled person. I only hope that in their attempts they create better interfaces for disabled people. Perhaps even get people waking again. For people who can already walk and use a mobile phone I doubt a brain implant will be much use. It’s hard to beat eyes and ears for getting information in.

That’s supervised of a different sort. Also, it may end up learning the effects of the noise which may not reflect real data.

It says you inhabit a different hierarchy.

Status has been important to humans since before humans were humans. It used to be a matter of life and death.

I believe the only thing that has changed is our perception about our place in the status hierarchy.

Mass media has the effect of making people aware of more high status individuals and this has been exploited by advertisement. But people on TV are easy to keep in a separate status box.

The recent phenomenon has been social media and the positively biased posting by peers giving a false impression of low relative personal status. There is also a bias to pay more attention to those of higher status, the more people you know the more individuals of higher status you’ll know. Both of these effects can be undone by understanding selection criteria biases.

I think the solution is to teach others of the effect of selection criteria biases.

There is also the solution of breaking down the hierarchy into smaller niche hierarchies so more people can be closer to the top of their smaller hierarchy. E.g. maybe I’m not the richest person alive but I have one of the rarest sneakers, or the best Magic The Gathering deck, or Furry costume etc.

I think there is a comercial drive towards niche activities that are now more viable with targeted advertising. So the disease may become the cure as there is money to be made in taking away a small subset of a bigger status hierarchy and creating a new one.

Consider this as opposed to a communist utopia where there is only one hierarchy.

Google lvl 5 total comp is ~$320K. So it is pretty close and still less than PHP consulting.

I used to work for them. I still found that I have a $300K ceiling.

It’s face recognition software, I don’t have a website, and for obvious reasons I’m pretty quiet about it.

The field has changed so much that I don’t know if I could still give advice.

Initially one of those elite consulting marketplace things. And then my own pool of referrals. I hate PHP so I set my rate higher on those projects and people still pay it.

I’d have a hard time finding a job in ML for $300K but I’d have an easy time finding a job in PHP contracting for $350K. This is despite having 10 years experience as a ML practitioner.

I quit my last job after bringing in over $100M in profit and only getting a $30K bonus.

Now it would take ~$500K and an ‘eat what you kill model’ for me to consider going back to work.

These days I make my money from licensing ML products which gives me a lot of time to study - which is what I enjoy doing.

That’s by design. If they had a good search people would discover content was missing faster.

An open question for me is the performance of two 2080tis using NVLink as one virtual GPU. I imagine it’ll be close to linear, but I’ll be interested to know for sure.

While we’re at this, has anyone got significant training perf improvements for Titan V. Best I’ve seen is 2x of 1080ti for 4x the price.

Deep Learning, once you’ve got enough GPUs you end up bottlenecking on the CPU.

I built a face recognition algorithm to sell to police, immigration and pretty much who ever wants it. It took 6 months and is state of the art. Which is to say, not as good as a human, but getting close.

If you’re worried about privacy then maybe you should take the tracking device out of your pocket.

The concept of using the power of the state to limit the power of the state seems flawed to me. Especially an oppressive state.