thedragonline

What are you talking about? I spent years munging climate datasets from various research institutes around the world. The upshot: it doesn't look good for humanity. I seriously don't understand how a neutral third party can walk away from this climate work and think, 'Nothing to see here folks.' The denialism and willful ignorance of the potential catastrophic consequences is something I find terribly disheartening. Mark my words - the temperature records that keep getting broken year after year are going to keep getting broken. Entire towns going up in flames and cities being wrecked by increasingly more powerful hurricanes will be the new normal. <sigh>

Yeah it did. I'm just sad that it took down ActionScript with it. IMO this is the language Javascript should have been.

From the mouth of John Roberts himself: "courts must exercise their independent judgment in deciding whether an agency has acted within its statutory authority." This is not going to end well.

As an aside, I've always wondered how society will drive the response to climate change into a ditch - it'll be at the hands of lawyers (because, you know, they are policy experts).

>People who have worked in the second type of environment (deadlines are real and have consequences) become quite good at estimating and managing progress toward deadlines without crunch mode at the end.<

I don't know. In my experience, estimation is one of the most difficult aspects of software engineering. A black swan event or some unknown unknown will blow apart your thoughtful estimate (especially near the end of a development cycle) - no matter how good you are at it. I'd much rather have a crystal ball at hand, and let the chips fall where they may, than engage in a herculean struggle to accurately predict the future.

edit:formatting

There is work to do in this world that offers no remuneration or very difficult paths to said remuneration.

Work must always revolve around a pecuniary interest at all times?

Anecdata: I tell a relative that climate is going to get hotter and more variable and that said relative should be concerned that an unaffordable electricity bill may be in the cards (Houston anyone?). Relative's response: the governing authorities won't let that happen. Me: <internally sighing and filled with despair>

I spent half a decade working with climate data from institutions around the world. There are dragons in those datasets and it is really frustrating to watch denialism at large and have to deal with it personally.

I'm actually not as opposed to belt tightening as you might think. I just get aggravated seeing narratives without a whiff of empirical evidence to support them. (I've informally studied economics and probably know enough to be dangerous.) My personal indicators that we may be on shaky fiscal ground are the rise of monopsonies/monopolies in vertical markets that depend on discretionary income (concert tickets anyone?) and the most troubling chart I've ever seen from the BLS (Bureau of Labor Statistics) https://www.bls.gov/bdm/entrepreneurship/entrepreneurship.ht... (scroll to the bottom and look at the last chart).

Let's set the rhetoric aside. What debt to gdp ratio would be reasonable? Do you have any sources to support said ratio?

OK. Let's take the U.S. out of the equation and look at Japan's inflation - https://tradingeconomics.com/japan/inflation-cpi. Hmmm... you'd expect with the 'unsustainable' government spending in Japan that there would be an uptick in inflation. There's nothing of the sort - unless my eyes are lying to me. And this goes back to my second point - nobody really knows what levels of government spending are sustainable, and my guess, given that it's not just Japan with a high debt/GDP ratio, that it's likely a much larger number than some posters here would like to admit.

>It’s so clear this isn’t sustainable. Maybe. Having that much debt is problematic in a household budget. However, governments are a different animal - just look at Japan's debt to gdp ratio (https://www.imf.org/external/datamapper/CG_DEBT_GDP@GDD/CHN/...). Posters who claim the US Gov's debt is strangling the nation are obligated to explain why Japan hasn't imploded. The likely real answer: nobody knows what levels of debt are sustainable for a government.

FWIW, I don't have kids. And given what I've seen in the data, don't intend to.

IMO, it seems these people exist to muddy the waters. I spent half a decade munging temperature and precip data from multiple institutions around the world (MPI, fer instance). These datasets suggest an unpleasant future, particularly by mid-century. But I'm supposed to ignore all of it because a) experts are wrong b) sunspots!

Let me contextualize that "some" for you: https://fred.stlouisfed.org/graph/?g=17rEg

Real wages have effectively flat-lined over the past 50+ years.

I'm not sure removing options from the desperate and dying is a good idea. Those with means will just ignore your concerns and the rest will have _nothing to lose_.

I miss AS. I used it to build a gis engine that packed 5D into what superficially looked like 2D - item renderers helped make that happen. It is aggravating to no end to watch people pile on Flash for its faults - they either don't understand what was lost or just don't want to understand the loss. Steve Jobs is in this group.

I wonder if gpt3 has a use case here?

Uh, then don't use it? Problem solved methinks.

Definitely. There is a weird elision on the author's part that other forms of artistic creativity exist. FWIW there is plenty of room at the bottom for those artists that can pull together the tenuous threads between seemingly incompatible disciplines.

Ex: A down arrow button attached directly to the right side of the year input field. Make it a deeper blue and then wrap a 1 or 2 pixel wide rectangle around the field itself of that same color.

>I think this is obviously false and hurts the credibility of the rest of this comment.

What you think of this graph? https://www.epi.org/productivity-pay-gap/

Because by my reading, it looks like the average American is not reaping the gains of this country's GDP output by a significant margin.

Yeah you should have. I was about to comment on how toxic this kind of worldview is, but held back. At any rate, this kind of thinking ensures a world where opportunity is increasingly siloed such that even the best and brightest wind up losing for want of wealthy benefactors.

Interesting. Do you have a concrete example that would illuminate the connection?

This quote strikes me as something quite apropos to your Father's difficulties: "In its majestic equality, the law forbids rich and poor alike to sleep under bridges, beg in the streets and steal loaves of bread." Anatole France.

I've already built a working model (at least it appears that way). I'm looking for any research involving non-representational, low resolution imagery (3x3) - I haven't seen anything that rules out working with this kind of data at this scale. But maybe it's a fool's errand. I don't know. It would be nice to find literature that catalogs what has/hasn't been done in this space, or better, what works and what doesn't.

I personally suspect that there are surprises in store for me by rooting around in this space. I have a .caffemodel that appears to correctly label better than chance, so my next step is to create a test dataset that's independent of original and see what happens. Personally, I find it unsettling that GPT 3 now contains 175 billion parameters. Why not go in the opposite direction and test the lower bound of what is possible?

The pooling layers are gone along with the first RELU. I've also reduced the number of outputs in each convolution layer.

>Have you run your test multiple times with varied training sets?

Some background - I've been working on this for several months now, experimenting with various CNNs (settling on a modified LeNet), hyperparameters and parameters. The bulk of the experiments have been failures - a typical scenario is the loss function decreasing in the training phase, but winding up unable to correctly predict labels in the test phase. There has been a progression however - from predicting either label A or B (but not both), to predicting both (but no better than chance), to doing a little bit better than chance. Maybe I'm fooling myself - I don't know. That's why I've been scouring the Internet for similar kinds of work (and not finding anything truly useful) and now reaching out to HN. If there's work out there that definitively rules out working in this space, I'm all ears. Otherwise I'm going to keep on experimenting. (edit:formatting)