benp84

Those numbers are in two different units: wealth is $ and GDP is $/time. Increasing wealth/income ratio could mean people are saving more.

As a combinatorial optimization problem, integer programming is the best tool for solving Sudoku. Fast, accurate, and guarantees the optimality of the solution. I'm impressed at how well this neural network did anyway though!

Well, I think I found the problem: despite my meticulously-defined bucket access policy, it turns out I had write permissions enabled for "any authenticated AWS user" in my access control list. I did not realize there were two separate pages for these settings.

I suppose it's possible that the bot enabled this setting, but it was probably just me being sloppy :-/ The bot probably scans for poorly-protected S3 buckets that are referenced on websites.

I hope the next victims find this post in a Google search.

That's my impression too, because no one would manually add plain text JS to a .gz file, but wouldn't that imply that someone has broken into so many AWS accounts that they needed a script to hack them all!?

The tracking code is definitely in the file. The checksum has changed, the last file modification timestamp is well after the deployment time, and I can see it when I download it straight from the S3 API.

Yes, sitewide SSL. The site is [redacted].

Affected files are:

https://[redacted].s3.amazonaws.com/js/scripts_2017-05-28-17...

https://[redacted].s3.amazonaws.com/js/scripts_2017-05-28-18...

According to "last modified" timestamps, the first was modified 7 minutes after upload, the second 2.4 days later.

"can generate a magnetic dipole field at a level of perhaps 1 or 2 Tesla (or 10,000 to 20,000 Gauss)"

I'm just amused by this conversion. Who is this for? Are there people who know one unit of magnetic flux and not the other?

Tim Ferriss has "a little trick" to supposedly solve every complicated life problem in 2 minutes.

I think his comment was mostly figurative.

I wonder if any of these farming tasks would be more efficient with a swarm of mass-produced drones with the right software. Does farming currently use such large, expensive equipment because it leverages the attention of the operator, or because there are actual economies of mechanical scale?

I once built a MS Excel add-in in C++ that worked perfectly when compiled by Visual Studio in debug mode but would not run when compiled in release mode. Obviously, debugging in release mode was a huge pain.

Long story short: at some point Excel stores the file path of the add-in in a string with (for some reason) a max length of 128. The file path "C:/.../Debug/addin.xll" totaled 127 characters and the path "C:/.../Release/addin.xll" was 129.

Don't forget the DE annual report fee ($50) and the DE annual franchise tax ($175 - $180,000, depending on share count). Plus the registered agent ($50) I pay $275/year.

[https://www.delawareinc.com/delaware-franchise-tax/]

The bigger cost for me is the C-Corp tax returns, average cost $806/year, but I suppose that applies to C-Corps everywhere.

[https://www.irs.com/articles/tax-preparation-costs-and-fees]

That's per second of average transaction time, meaning one second times the total number of transactions (per year I guess?). Kind of a meaningless metric in my opinion.

Great resource! Thanks for the link.

Subscription pricing is also a mechanism of price discrimination: users who need it once pay less and users who need it regularly pay more, as they both should. The author is fortunate to be in a position of sustaining a business on new users alone, but most companies don't have the margins to afford to leave that much value on the table.

Except that one requires login.

Did you read the article? They're basically A/B testing their notifications and forms. It's nearly costless and they quickly scrap what doesn't work.

So in other words, a bunch of people guessing which hypotheses were true was more accurate than actual scientific studies of them (71% vs 39%). Great.

That was my take-away too.

They had an idea to fix a problem, it worked but not as well as hoped, so now they're learning and iterating. Sounds like any experimental project (or startup).

They got 28 million stoves out with a budget of $10m/year (times 5 years?). Big deal if they haven't saved the world yet.

Light moves 3*10^8 m/s in a vaccuum, which is 7.5cm per clock cycle of a 4GHz processor. The processor is about 2cm wide, so not much room for improvement I figure.

I'm pretty sure it was a comment on the state of the education environment, not a personal opinion.

I don't expect there to be much debate about conditions like "they gave you their email address on your website and your email has an unsubscribe link."

It obviously wouldn't be a perfect system, but neither is any reputation formula. And the consequences of being wrongfully blacklisted are incredibly mild: I'd just skip our follow-up emails, or maybe require blacklisted users to double opt-in.

Hack the gas station by driving off!

Great idea. I'm not sure about the pricing though. I'd expect anyone who reaches 1000 submissions per month to probably make the effort to get their own form handler. I'd make it 100/month or maybe 1000 submissions total until you have to pay. Then make it an annual fee so that customers think "OK, I'll pay $X not to have to deal with this for another year."

This is a great idea. You should build it!

I await the day when a single article is published with a proper unit of power or energy. If it's not a 54 kW battery it's a windmill that generates 3 MW per year.

Keep in mind that users are logging on and off all the time, so it has to be an online algorithm.

I can't think of any other way off hand. I'm assuming you don't like cells because of edge cases where two people could be near each other but in different cells? If the cells were carefully selected it might not be too big of a problem. For example, you could put each college campus in its own cell. But that could take a while and you'd have to drop the one-mile guarantee. I'll let you know if I think of anything else. I don't think it has to be a dealbreaker for an otherwise great idea!

Congrats on the success! I'm curious: the condition of <1 mile proximity isn't transitive, so wouldn't users be seeing only one half of a lot of conversations? Like if A was close to B, B was close to C, but A wasn't close to C, then A would only see what B says to C but not what C says back. Or have you divided up space into fixed cells roughly one mile wide to prevent this?

I got a job offer from Google that started from them finding my LinkedIn profile. That said, I despise them more than any other major website. Their entire business model is holding the information for ransom that you provide them to make available to the business world. I'd love to see a YC company replace them, or do so myself.