massaman_yams

1. Identify a large number of email newsletter signup pages that don't use captcha and which either send an opt-in confirmation message or a welcome message on signup.

2. Identify a target for some kind of account takeover attack. (Assuming you have other details needed for takeover.)

3. Rent botnet.

4. Perform thousands of signups for the target's email address starting shortly before your attack.

If the account's only security notifications (e.g., password reset, etc.) are in the form of emails, the flood of spam will usually keep the target from seeing them until too late.

These are real attacks, frequently seen in the wild.

Interesting conclusions, but small sample size: N=20.

Vitamin D at normal doses is quite safe, though, so I'll be supplementing my intake now even if broader long-term study shows this to be ineffective for C19.

No. My original comment wasn't worded well, sorry. See my clarification above.

A handful of friends/acquaintances on fb - so, no, sorry. It's clear they're less informed than most, and I've certainly seen plenty of more appropriate responses along the lines of what you describe.

To clarify a bit: it's certainly not the position of the opposing party as a whole, but those I saw who were vehemently opposed to it were ideologically motivated.

Politically, yes, there was/is a cargo cult. Both for those convinced it's a cure, and the anti-cargo-cult opposing party convinced it's somewhere between useless and actively harmful.

From a medical practice standpoint, you have a widely-available drug with well-known side effects that can be administered in a medical setting with low risk, and potential life-saving upside. When the research is early but potentially promising, it's not irrational to administer in cases where the side effects should be well-tolerated, even if later research shows no or low benefit.

The problem arises from self-administration without medical supervision, or with inadequate medical supervision.

The URL is a little misleading here; when scanning the frontpage it appears the project is associated with Google, when it's really just a Colab notebook.

Maybe the Github repo would be a bit clearer to link to: https://github.com/firmai/deltapy/

Wasn't clear you were saying it uses copy just like pgsync uses copy. I read it as you making the distinction between the two.

This uses COPY for export & import:

https://github.com/ankane/pgsync/blob/master/lib/pgsync/tabl...

One-way, yes. pg's COPY command can't do upserts, this can.

The primary use case for this appears to be ETL. Pg_dump is more backup-oriented and optimized for larger operations, vs. a bit more fine-grained in most ETL processes.

It's batch sync, not streaming or replication. By default, appears to dedupe based on PK.

I send between 500 and 1000 per month through mailgun. On high-usage months, I will pay around $1.

Even if I sent 5000 messages one month, I'd pay around $4.

This is not even close to a significant cost.

No. Botnets are large and broadly distributed enough to render protection methods based only on the IP or IP block ineffective. They're commonly used for mailbombing attacks such as those described here: https://www.wired.com/story/how-journalists-fought-back-agai...

Do you think a botnet with 10k machines is going to be meaningfully inhibited by making each machine's cpu run calculations for a second or two for each submission?

I'm sure reCAPTCHA looks at the IP and IP block as one of the inputs to its ML algorithm, but as one or two of perhaps a dozen different features - including mouse movement and/or keyboard input, which is quite a bit harder to fake.

One thing known to speed alcohol metabolization is fructose. That's not to say consuming fructose will make you "sober", though, as the first metabolic byproduct of alcohol - acetaldehyde - also has CNS effects.

https://www.ncbi.nlm.nih.gov/pubmed/1854373

Or, you can look at the (say) 6 month sales cycle for an enterprise platform, and the internal political wrangling that may be required to get approval for the line item, as significant hurdles to moving forward with an ETL project. There are some legitimate, and some less valid, reasons for many engineers' (and engineering-driven orgs') bias against commercial options; and yes, sometimes this results in over-engineering.

Airflow can do some things many commercial tools cannot, though, so for some it is the right option.

Right, AutoML != AGI.

Built-in regularization, probably, plus cross-validation. These techniques aren't new; they're included in a number of ML libraries already - just not at this level of automation.

I did it with a Pi, HomeAssistant, smart bulbs, and some Python, but it was not as good or as reliable as I'd like.

I wanted the lights to know what temperature and brightness to turn on to based on a schedule, but there wasn't a way for them to pull the desired state upon power-on; instead I'd have to detect power on and push a state change to the desired color temp. This meant they'd turn on at whatever last state they had on power-off, and a few seconds later (or frequently much slower) adjust to scheduled color temp.

This may have worked better had I used smart switches to turn lights on/off rather than wall switches that cut power to the bulbs, but I also had issues with the smart hub losing connection to bulbs occasionally, HA not reliably seeing power-on events for the bulbs, and HA losing connection to the hub. (Bulbs/hub were Ikea tradfri; perhaps other manufacturers' products are more reliable).

This was a couple years ago; I haven't looked into it much since - maybe it would be easier to do now.

Sending text-only is probably a weak spam signal as far as Gmail is concerned; for the majority of mail to the majority of recipients (though yours may be exceptions), HTML email - even if very basic - offers better layout/design options and better readability.

On top of that, of very old-school users on Mutt or Pine who prefer text over HTML, I'd think a significant portion of them don't have their mail pass through Gmail's filters.

And if you read the rest of that mailop thread, you'll see that the broader IP reputation of the /24 where your IP is located is a key part of whether or not your email gets delivered.

I would venture to guess that a substantial portion of people reporting issues delivering at small scale are sending from a low-reputation netblock. (Assuming auth is all set up optimally.) So it's not that it's not possible to send at low volume, it's just that most people don't know to look at the netblock, it's hard to evaluate netblock reputation, and it's not always possible to select a preferred netblock within a given provider (especially VPS providers with low-cost options).

The bottom line is: it's hard to run your own mail server, and much of the reason for the success of services like Sendgrid is a) it's easier to set up auth properly with most of them, and b) they actively manage the email reputation of their netblocks (though some do this better than others).

If you're approaching things from a practical perspective instead of making this your hill to die on, you just use an email service provider, and for the vast majority of senders, life is a little bit easier. There's plenty of ESPs to pick from, and there's a very good chance you can find a free one that delivers better than an average VPS or IP of a residential ISP.

Sometimes fractious? Ha. There's so many cranks there that 95% of the messages are not worth reading, even if most are well-intentioned.

Let's assume he's on the spectrum for the sake of argument. If you miss some indirect signals, that's one thing - but when you're explicitly told specific behaviors are inappropriate, and you keep doing them anyway, being on the spectrum is not a defense.

Labeled as 5% but actually containing 40%, sure. Labeled as 35% but containing 40%? No, because those are functionally identical for most consumers.

Clearly you don't have the full picture. This isn't about a difference of views; it's about a long-standing pattern of inappropriate behavior, including (but not limited to) overtly and deliberately staring at women's breasts in a professional context. There's no innocuous explanation here; this isn't an isolated mistake.

I was thinking along the same lines; but there's a potential problem in that this provides an incentive to mislabel a beverage's ABV to get a lower tax rate.

That's probably solvable-ish by fining companies found to have an ABV off by more than 1% or so, but that's a lot more work to monitor, especially with the number of different products on the market.

From the report linked to in this commment https://news.ycombinator.com/item?id=20833229

"There is moderate evidence that cannabis or cannabinoids are effective for: • Improving short-term sleep outcomes in individuals with sleep disturbance associated with obstructive sleep apnea syndrome, fibromyalgia, chronic pain, and multiple sclerosis"

That's not everybody, of course, but it's enough to say that weed can have positive effects on sleep quality.

I use the same helmet when commuting and on longer/faster rides.

yep - using the pg_trgrm extension. https://www.postgresql.org/docs/current/pgtrgm.html

Only if you're biased towards accepting those kinds of arguments already. There are social explanations for the effect that, in my opinion, both better explain it and don't suffer from the slightest appearance of racism.

Thanks for sharing! This looks much more practical than HoloClean.

It's slow, first and foremost; while I'm not 100% sure on the internals, I think that's because it's doing these operations on js data structures in browser, so pandas would be up to a few orders of magnitude faster out of the box.