HN user

dave_sullivan

3,925 karma

Email: dave.brian.sullivan [at] gmail.com

Posts19
Comments827
View on HN
davesullivan.is 5y ago

My startup failed, then I found out I was unemployable

dave_sullivan
460pts595
github.com 7y ago

Agile Lite: Agile without all the burnout

dave_sullivan
427pts280
davesullivan.is 8y ago

Why Elon Musk Is Wrong About AI Safety

dave_sullivan
4pts0
www.usatoday.com 10y ago

A third of US corporate cash is held by 5 tech companies

dave_sullivan
2pts0
www.marketwatch.com 10y ago

Here’s all the money in the world, in one chart

dave_sullivan
7pts0
xxx.lanl.gov 10y ago

Neural Programmer: Inducing Latent Programs with Gradient Descent [pdf]

dave_sullivan
59pts21
thestack.com 10y ago

New photographs from dead photographers with convolutional neural networks

dave_sullivan
1pts0
arxiv.org 11y ago

Weakly supervised memory networks [pdf]

dave_sullivan
1pts0
arxiv.org 12y ago

Deep learning in neural networks: An overview

dave_sullivan
1pts0
wordcloud.ersatz1.com 12y ago

Show HN: 3d Word Vectors with Neural Networks, T-SNE, and WebGL

dave_sullivan
30pts19
allfacebook.com 12y ago

Deep Learning: Facebook Wants To Know Even More About Its Users

dave_sullivan
52pts26
www.blackcloudbsg.com 13y ago

Ersatz neural nets try day trading

dave_sullivan
39pts11
www.ersatz1.com 13y ago

Ersatz - Deep neural networks in the cloud

dave_sullivan
121pts31
www.guardian.co.uk 14y ago

Brain implant allows paralyzed woman to control robotic arm

dave_sullivan
2pts0
www.sciencespacerobots.com 14y ago

Maybe there's life on mars after all

dave_sullivan
1pts0
www.businessinsider.com 14y ago

Groupon app disappears from appstore

dave_sullivan
3pts0
www.blackcloudbsg.com 14y ago

Those "Oh shit" Moments

dave_sullivan
1pts0
techtalks.tv 14y ago

Building IBM Watson: talk from ICML 2011

dave_sullivan
4pts0
www.cs.cornell.edu 14y ago

Detecting Opinion Spam w/ NLP

dave_sullivan
2pts0

My statements based on high level meetings I had at the time with all 3 companies when I had started an early neural network PaaS company and was looking for them to invest. Nvidia knew what they were talking about and were already moving that direction, Intel heard about deep learning somewhere but didn't believe there was anything there, and AMD didn't know anything about anything.

I'm personally not a fan of NVIDIA's drivers or the reliability of their hardware.

As compared to AMD or Intel? I wish there was real competition to Nvidia but there isn't. I'm not a fan of their defacto monopoly but they do have the best product on the market and their competition has been asleep for 10 years. AMD and Intel barely knew what deep learning was 10 years ago (and certainly did not appreciate the opportunity) and Nvidia was already investing heavily.

"Complete understanding" isn't a real objective. Understanding more than we knew before is more realistic, and yes, I'd say we understand more than we did before, if you take a sample every 5 years or so. Every 5 years is better and better. People keep working on it and always will. The alternative is giving up and stopping all research I guess?

I suppose it's impossible to prove the specifics of the conspiracy, but that Jack Ruby, a guy running strip clubs for the mafia, takes it upon himself to publically shoot Lee Harvey Oswald kind of proves to me that there was some kind of conspiracy. What other explanation could there possibly be? What Oswald might have said if we were kept alive... no one will ever know. But if there was nothing there, why did Jack Ruby kill him? Temporary insanity, deeply patriotic fan of JFK, or tying up loose ends on orders -- what makes more sense?

I do the same thing, don't hook it up to internet, never update, don't use smart features, watch everything off a separate Kodi box.

But TVs have become so hopelessly enshittified I would support legislation at this point. I am sure they are making a lot of money serving these ads ensuring the behavior will only continue and no "normal tvs" will even be sold.

Really good post and great example of how someone can incorporate these tools into their work. Maybe there will be a new resurgence of proc gen games that finally manage to get past the same-ness problem with more diverse, interesting generated environments.

Get Ableton or FL Studio or Reaper for DAW, get Dexed as a free VST (or serum, not free), and search YouTube for "how to make synthwave with X" and you'll find a bunch of stuff.

The DAW is software that organizes all your instruments and melodies and so on into a project file and a VST is the instrument itself, more or less. So you work on a song in the DAW using VST instruments.

"Not much" has fundamentally improved since Alexnet???

Honestly half the comments in this thread boggle my mind. What is even the word for someone that is shown something amazing that they don't understand and then dismisses it on completely superficial grounds?

Like if I go back in time and show someone a computer and they think it's just a typewriter and who ever needed one of those anyway? It's like ignorance but in the most willfull, dismissive way possible. Gross.

Don't be negative about revolutionary new technology that can make the world a better place.

As it stands, people believe whatever dumb thing they read online. Thanks to AI, they will have to learn that if something is out of the ordinary, they should confirm with multiple sources. I call this a net win.

And for those literally medically incapable of this level of reasoning, we will soon have "AI firewalls" (Gibson ICE) that can tell people "Hey, this looks like a scam!" and also help them reason about complex topics.

This is fantastic, what a time to be alive!

Netflix and other tech stocks did not lose their value over decades, it happened over the course of months. But yes, market volatility is a thing for stocks as well as bitcoin. They're the same in that regard.

"Bitcoin to zero" is a matter of faith at this point, I don't see it happening and I haven't heard anyone explain a good model for how it happens. US is certainly trying its best to outlaw it. Still won't go to zero.

I think my larger point is if the government wants to go after something, go after fraud--in crypto or "Fintech" or banks or stocks or whatever. And mostly they are! But then they're also leaning on banks to cut off the ability to convert crypto to fiat and that's a bunch of bullshit.

I dunno, Bitcoin loses 65% of its value and it's a scam, but Netflix does it and it's... not a scam?

SVB implodes and investors (but not depositors) lose everything and it's ok.

You're nuts if you think there's no insider trading or pump and dumps on the stock market.

Seems like a double standard.

This all has more to do with maintaining a monopoly on the financial system and certain classes of participant in the US wanting to maintain the power to collect rents. And there really is no counterbalance to their unchecked authority.

If heroin was legal and pure, there would be far fewer fentanyl overdose deaths.

The number of deaths from not-fentanyl is low so calling it "drug overdose deaths" instead of "fentanyl overdose deaths" is disingenuous.

We shouldn't learn anything from fascist Singapore.

Does this position make me a hypocrite?

A little bit, but it depends on your personal reasons for supporting legalization. You sound like you are more concerned about bodily harm of a substance, hence marijuana is ok because it's less harmful than alcohol (which we have accepted as a baseline "well of course alcohol is legal!", so we can then allow everything "less dangerous than alcohol")

I support drug legalization more from a "personal freedom of adults to make their own choices" POV. I'd even stretch that to include choices that statistically add strain to the medical system, like motorcycle riding or eating too many unhealthy foods. People shouldn't have the right to tell me what I'm allowed to put in my own body, and there also needs to be commerce allowed to service given needs, so defacto legalization of usage is not a great solution but is a start.

But I've been beating that drum for a long time, argued till I was red with many people who disagreed with marijuana legalization, now it's broadly accepted and those same people forgot how strongly they felt. "People should mind their own business" is a slogan I can broadly get behind and it applies to my opinions here.

Do you have any leads on short term contracts? Would accept one in an instant.

The problem is upwork takes a 20% cut and has bad terms and has a very flaky client base, while everything that VC money touches pays way better but hires non-contract fulltime only (fine) and has these rediculous interview cycles that I never seem to pass through.

I think if you put me in a stack of candidates that did leet code, i am basically never in the top 3 fastest, so that makes me unemployable in a system where resumes dont matter, personality is suspected to be bullshit, and speed of test completion is the only metric that can be used to rank candidates. Not in the top 3 fastest test takers? Get fucked.

Maybe the time is right for anti-agile coaches to call all the companies that bought agile coaching over the past 15 years. I guess they'd be talking about why waterfall was best all along.

I think some of the root issue is the annoying micromanagement and lack of trust from non technical stakeholders. They need to chill out, but it's in their nature to be dicks somehow. They think if only they plug numbers into a spreadsheet correctly, huge savings will just pop out. But really they waste time and annoy the hell out of people. Maybe anti-agile is "fire all the nontechnical managers".

Agile as a system of prioritization isn't bad, but human factors mess it up a lot.

Facebook released over 100 pages of notes a few months ago detailing their training process for a model that is similar in size. Does anyone have a link? I can't seem to find it in my notes, googling links to posts that have been removed or are behind the facebook walled garden.

But I seem to remember they were running 1,000+ 32gb GPUs for 3 months to train it and keeping that infrastructure running day-to-day and tweaking parameters as training continued was the bulk of the 100 pages. It is beyond the reach of anybody but a really big company, at least in the area of very large models, and the large models are where all the recent results are. I wish I was more bullish on algorithm improvements meaning you can get better results on less hardware; there will definitely be some algorithm improvements, but I think we might really need more powerful hardware too. Or pooled resources. Something. These models are huge.