time limits put an upper bound on how long youre spending on it which is good for a lot of people. ive given take home tests and some people clearly put more time on it than we expected.
as an applicant, ive refused takehomes that were auto-sent on receiving the application, but if its after a phone call with the hiring manager or engineer its not so bad. but then, i did end up having to defend why i didnt do a bunch of parts ancillary to the assignment - 'why didnt you build a config loading system?' 'uh cause i timeboxed it and that gives you fairly little signal compared to the business logic'
"5. Your PointCard Titan spending limit is based on your net assets. As your net assets grow — in the form of cash, investments, cryptocurrency, etc. — so does your spending limit."
i had a job where i was the first staff, but when they started promoting everyone to staff cause every senior wanted a promotion, they had to create new levels of staff to differentiate
it ends up being cheaper overall if you have high utilization of your provisioning since the per second fee while a function is running is cheaper. using https://calculator.aws/#/createCalculator/Lambda, if you have a steady 1 request/s and each requests takes 1 second, 2592000 seconds in a month. at 1024mb, i get 36.52 for provisioned and 43.72 for on demand. With autoscaling...you wont get 100% utilization, but it probably ends up being close enough to a wash
they exist, but arent really used. the us is much bigger / less dense intercity. going across the coasts is six hour flights but days(4ish i think) by train and significantly more cost, with not many cities in the middle that people stop at. too few people would use so theres no motivation to make it faster. where it works decently well is northeast, because the cities are very close together(dc,philly,nyc,boston). but it's usually much more expensive than driving/bussing and takes roughly the same time.
Games are especially bad. They are relatively underpaid and overworked because lots of new grads come out of school and want to work on games - the cynic in me would think they’re digging at how much work they can extract from you. The last straw for me at one gaming place was leaving at 7pm and someone commenting about leaving early.
I don’t think that answer was terrible, but could be tweaked a little. Go into what technical challenges you think that company would face and why it excites you, but don’t present it as a negative and I wouldn’t blink an eye at it. but at the end of the day, youre not there to solve cool problems, youre there to make money
That said, I worked on iPhone games that nobody wanted to play at the company. One day our metrics tanked and people were spending hours trying to figure out why. I just opened it and in thirty seconds the first screen related to those metrics was blank. Wanting to dog food the product, if anyone actually had used it, is very useful
does blockchain solve the trust problem here though? my trust issue would be that the person doesnt report their entire income, or purposely delays deals past the term, which is all out of band of the coin
We've had to deal with a few. They removed a gcs api a couple years ago and told you to update to the newest version of the sdks. obviously they didnt tell you which version it was actually fixed in, figuring that out was an adventure in itself. everyone's runs off master right? Worse, they hadnt even updated all of them by that point.
We updated to working versions, but support kept telling us we were using the old one. Turns out they didnt update the dataflow runtime, which they own and we cant inspect or update, and they were reporting it to us for us to fix...
"This action creates a copy of the object with updated settings and a new last-modified date. You can change the storage class without making a new copy of the object using a lifecycle policy"
"This action applies to all objects within the specified folders"
I'd assume the first one means youll get charged once for standard and once for glacier
I don’t think it’s purchasing power so much as risk pool. Or really a bit of both, it’s number of healthy vs unhealthy people in the pool. Employees of a company tend to be healthier. People on ACA tend to be part time / low wage, and have more health issues. Insurance companies pass that through by charging based on risk. I imagine what would happen is this would average out, the total cost(including employer contribution) would go up for most of the hn crowd who can afford it, but down for others who currently can’t
i imagine twitter has more in absolute terms, just due to the relative size of the two sites. I think the major difference is worse than just refusing to remove it, Parler advertised itself as the place where you can say things that most sites would moderate away, it actively encouraged it, so as a percentage it was much larger
I think both could be true. Yesterday it was just aws, and maybe they had a hosting provider lined up. But then today it became every vendor that they work with which is another level of problem, like having to build their own sms system.
Working in iPhone games, our biz dev people made us integrate with the shadiest of ad publishers. But we required attribution and could track how much the users spent, so could stop spending on them quickly, which was the normal result. Most of the companies were surprised we wanted attribution and had to build something custom, which to me means at the time nobody else was tracking it so lots of customers must have been getting fleeced. From the tweet, it’s crazy that they were spending 100m without tracking the quality of the users.
Also worked in ad tech. There were some types of customers that cared about performance, but the largest were typically agencies and all they cared about was spending the budget they were given. Everybody knows this is happening, you can differentiate yourself slightly by doing better detecting it. But at the end of the day people really care about customer acquisition cost which includes the fraud, so if you had 0% fraud everyone would just raise their prices or margins
Yea, it’s that smart, but it’s a risk mitigation cause people make mistakes. You don’t really want the thing that casually you run many times a day to have the ability to accidentally delete your database. Security as well, our code deploy pipeline has less privileges than our terraform pipeline
For sure on how exploitative it is. Just frustrating that so much focus is placed on ensuring minimum wage for these drivers but not those drivers. But if ride share minimums end up more expensive it seems kinda like proof that taxi drivers can’t be making minimum wage either, which especially sucks since government sets their exact prices
From skimming the bill, they assume you’re at 50% utilization, so for every minute you have a passenger, you must get paid at least two minimum wage minutes. So it doesn’t matter which platform or if you’re idling.
I think think this is the actual bill. So it is minimum pricing on time and miles backed into by hourly wages after utilization.
So 1.33/mile, but I think the time is total time not idle time like taxis, if so assuming an average of 30 mph add another dollar which puts it right at taxi pricing, and probably doesn’t include ubers fee.
i can't find the actual minimums, this[1] article suggests that the wait time is 0.56/min. I'm curious how this ends up comparing to taxis, since a taxi has a idle time of 0.50/min in seattle - it seems unreasonable to set minimums higher than taxis(but maybe it wont since i cant find the per-mile fee, and 25% over the current uber per-mile fee is about $1 less than a taxi)
This is the argument aws uses. In a dc you might need a bunch of high price Cisco equipment, and that’s part of what is getting priced into the bandwidth.
the cynical part of me agrees though that making moving out difficult/expensive is a large part of why too. Probably some combination of both
i dont think most companies are talking about ending the office culture. I know for sure my company is not, and its one of the ones thats already publicly stated WFH is a permanent option. There are a ton of people like you who would rather work in an office. There's also a ton who would rather work from home. It seems like the expectation is more work wherever you work best, home or office or hybrid. That does mean you will probably have more remote teammates, however I expect there will be density in offices for a long time still.
If I recall correctly, a couple years ago aws switched from one central team owning all of the cfn implementations to pushing it onto the service owner, so it’s no longer bottlenecked on a single teams queue. Given that the serverless teams are all invested in sam, it follows they’d make sure the things are implemented
I used to do ivr software. There really is that many people calling in. At least, ten years ago there were, I’m sure there’s been some shift but I don’t think it’s nefarious
Ah I meant that as separate comment, not hbase specifically, but that data pipelines need updating over time. Generally since what’s worth optimizing for changes as size increase, and there’s always that years old pipeline that takes hours to run that with a few changes could be minutes
I’ve had some bad luck with hbase for similar but not tried spark - it’s better optimized for writers than readers iirc. Pinterest[1] did an interesting thing where they worked directly on the hfiles. Everywhere I’ve worked I feel like I could pick any random data processing job that took more than a few minutes and get orders of magnitude improvement, there’s just so much inefficiency
At least when I did similar, moving the data itself was not the problem - it is entirely s3 server side. The problem is round trip times on the empty-body api calls themselves, aws cli being python is slow and maxes out on https/signatures.
If the copy Api supported wildcards there wouldn’t be any discussion at all on this
I had to do this before batch, replication, inventory. Support, SA, TAM were all like... good luck, though they did offer to make an inventory but they delivered that two weeks later. Everything at the time said s3distcp was the only option. It honestly wasn’t that bad, except it did a serial list before it started copying which would have taken days so needed some tomfoolery.
hah yea, their sales pitch here left a bit to be desired. Once a user gets a low score a recaptcha implementer should make a cell phone number collection system to have their own trust system based on that...