HN user

brian-bk

38 karma
Posts0
Comments49
View on HN
No posts found.

I think it's time that AWS customers hesitate before using any abstract software that AWS has built, as we're about to see the enshittification/profitability squeeze of an incumbent player. I wouldn't trust anything more than the ec2, RDS, S3 layer which are more "completed" products and there's not much to enshitify, only squeeze costs. You can already see that AWS has essentially dropped support of any publicly available software, I can't imagine it's any different internally.

It seems like AWS is replicating GCP's level of customer support, instead of using that as a differentiator (not that AWS' support was ever great, but dang it sure was better than GCP's by comparison). Consider yourself on the own, only trust software built on VM's, not SaaS abandonware.

I'm convinced nothing solid has come out of AWS in 10 years. The original products (ec2, RDS, s3) are solid either from being created at a time when AWS could write software or just from existing for so long.

Any modern AWS product (mostly SaaS) is just utter dookie, the only reason they continue to sell well is companies already have an accounting relationship with AWS. It's a lot less internal approvals required to start using the AWS SaaS version of whatever, compared to a brand new vendor negotiation contract and security guarantees.

So I guess AWS is just working off of a broken incentive structure designed to churn out SaaS-slop over stuff that actually works great.

I'm pessimistic about Go if they _don't_ change error handling. The handling of nil's and errors (by extension panic's) is purely conventional, and baked in to the language has no protection.

Author asserts that error handling is unsolved and therefor a solution shouldn't be attempted. I disagree, anything is better than Go's error handling now.

Investors love a company spouting AI not because it's actually good at raising efficiency, but because it's a codeword for layoffs.*

* there are certain industries that will see incredible efficiency gains from AI, but I think most will not see the efficiency gains that AI companies tout.

JSON(newline delimited or full file) is significantly larger than csv. With csv the field name is mentioned once in the header row. In JSON every single line repeats the field names. It adds up fast, and is more of a difference than between csv to parquet.

I'm more familiar with kubernetes for running production-grade apps and that's what I went with. I run ubuntu's microk8s, and the volume management is in kubernetes-land which I understand well. Finally helmfiles deploys my Postgres and immich helm charts. Perhaps "overkill" to some, but it's tooling I'm very familiar with, and feel comfortable running in "production" for my personal server.

Worth noting that immich itself recommends not using immich for the single source of your images and videos from how active development it is, so I also run nextcloud + crunchydata's postgres also inside microk8s/kubernetes.

This happened with me -- VP said he wanted feedback but clearly didn't like real critical feedback. Then that VP tried to set up a denied promo, even though I was performing two levels above my then-current level. Finally the company got bought and levels were frozen and he was laid off (as executives are often laid off in an acquisition), so who knows what would happened.

Gambling ads have become ruthlessly widespread, and I expect as much. They're way more common than alcohol or other advertisements that thrive on triggering addicts.

I really wish that federally the US came down harder on online advertising for addictive substances/abuse, it should be trivial to disable such ads. I believe that federally is the only real way to tackle online gambling because if NJ didn't allow these shady businesses, some other state would have and NJ's citizens would have still suffered, so might as well take the tax money you can get. (you can replace NJ with any other state that has now allowed online gambling businesses in the last decade).

I feel like the framing of this is pretty ridiculous -- employees are insulted by phishing tests that evoke an emotional response. Do they think real phishing emails would avoid such topics? Do they want the phishing tests to not actually train them for the real thing?

I thought this would be about tech customer service, which has gotten awful. Dark patterns and reduced human interaction galore.

Instead they talk about in-person customer service. I don't mind the more "realistic" personality, the fake happiness was always weird. As long as I can still get the item/food/service I need, I don't need a customer rep/waiter feigning fame positivity when they're just trying to do their job. To be clear -- nothing wrong with unadulterated true positivity, which is still great to see -- but if someone's clearly overworked and overwhelmed, it's okay to bring a more honest representation of their personal state as long as they're still doing their job.

My exposure is only anecdotal but I think there's been a significant increase in small scale and sophisticated cyber attacks since the war in UA.

I don't have evidence for why, I hope those in the media are looking into it. My total guess is I think Russia has implicitly decided to look the other way with hacking groups as long as they don't target "home". It'd be a win-win for the Russian govt and hacking groups: * destabilize USD * cause civil damage to citizens of 'hostile' nations (the US, from Russia's perspective) * fund further criminality * maybe even fund the war directly

Unfortunately many US companies are dumbing down their security orgs with layoffs, making them even more vulnerable. This will really be where companies shoot themselves in the foot. These layoffs could be another unique explanation into why I think there's been an increase in cyber attacks by increasing vulnerability, so maybe the reasoning is more innocuous than my guess.

Is it just me or is the issue of 1. Competition and 2. Security of companies using cloud providers just two separate things? There's ways to shoot yourself in the foot with security if you self-host or use the cloud.

Now the means in which companies get off pretty much with no retribution for security carelessness is a problem, but I don't know why the FTC would only take the cloud provider angle. Punishments for reckless handling of user data should be more severe regardless of whether a company decides to use the cloud or not.

I can see where this sentiment is coming from, in that there's very few objective markers (genetics, location) that can fully describe race. Further, I agree that studies should avoid generalizing and assuming that those markers are always there.

However I firmly disagree with the combative headline -- race is a very much a _real_ social construct. Just because it's a social construct doesn't make it any less real, identities and personas do have pretty strong experience's in one's social life which can impact their health, well-being, etc. I don't see why this opinion piece has to describe it this way, it minimizes the real impact these social experiences have.

Google's data suite (bigquery, ai) is unmatched imo but I agree -- prod realtime apps shouldn't run on gcp. Google has proved time and time again that they have awful support. I've seen plenty of gcp incidents and every support thread is them gaslighting that the customer is wrong, gcp had no issues. One time we had our dependencies on Google breaking and google suite reported the outage but gcp didn't, gcp support still tried to deny that gcp had any issues.

It's funny that AWS support isn't super great absolutely, but relatively boy their support is amazing. They will at least acknowledge there was an issue on their end.

I've found sorting by most recent reviews to be quite effective (works well for Amazon too from being burned on product quality cliffs). Recent reviews are a better sample than an average from everything, and give a better intuitive sense of what the place is like. IDK how some stays can have great ratings and the last 10+ reviews are all 1 star. Learned that lesson the hard way.