saryant
saryant on Hacker News
The issue there is running it on k8s. On bare metal or even VMs it’s not at all hard to run even at scale.
You’re off by a couple
orders of magnitude on what it takes to run Twitter.
If you don’t see how this strategy is tax inefficient then you have no business doing this.
Because your idea makes no sense, to be frank. The tax inefficiency alone would be absurd.
The 36 corridor between Denver and Boulder was home to a ton of storage tech startups in the first dotcom boom. There’s still a street called “Tape Drive”.
The API was 10% of the company’s revenue before Musk took over. That’s a good amount of value.
The v2 API was built over the last several years. It’s not new and was announced a long time ago, well before the acquisition.
Source: led the search API at Twitter until last year.
There is one SRE left from that team. That’s it.
You need both. Without the free API you don’t have a funnel of new customers who can grow into the enterprise tiers.
They have a hell of a lot to do with each other beyond that too.
(Led the search API team at Twitter and worked on both API platforms)
Showing ads isn’t enough. You need to count as impressions to bill advertisers and convince them to trust the impression data coming from third party clients.
Twitter made hundreds of millions a year selling API access.
We gave academic researchers free access to the full archive search API to help cover that use case.
Delivering old tweets is actually more expensive than real-time because nothing is cached. Streaming access is pretty much all served out of memory.
I used to lead the search and historical API team at Twitter.
Wanting peace is a good thing. Wanted Ukraine to capitulate to genocide is not.
Canon makes its own sensors. They’re the only major player not using Sony sensors.
Twitter distributes config files across their entire fleet (hundreds of thousands of machines) using git and a cron job. Commit your config change, wait a few minutes and there it is.
Source: I’m a former tweep.
It does matter, as it could be used to figure out user growth. Tweets and users now use snowflake IDs which provide increasing but non-sequential IDs which also encode creation time.
No he’s just making shit up. That statement of his is completely untrue.
Elon’s own data scientists agreed with Twitter’s bot count.
1L is the forward-most door on the port (left) side.
IFE is in flight entertainment.
I was on the ground floor of a modern building in Tokyo when Fukushima happened and barely felt a thing. These systems are remarkably effective.
Some regions operate TER trains themselves.
With a few small exceptions Vanguard runs index funds. They can’t just sell a stock because they think it’s overvalued, they must stay true to their chosen indices.
Sampling doesn’t imply synthetics. It just means that at the small end of their index they don’t bother to hold everything because the cost of doing so is too high. This is only an issue with niche ETFs or total market ETFs where the tail end of the index represents some tiny minuscule fraction of the fund.
Nobody uses sampling for, say, an S&P500 fund.
They’re bonds backed by the full faith and credit of the federal government. They’re as high quality an asset as you can get.
The problem is ensuring that event has been propagated all around the world so you aren’t displaying now-private tweets. That’s why we need strong consistency.
I have spent much of my time at this company on this exact problem.
Loglens, there’s a word I hope to never hear again.
I’ve done a version of this the last few years.
1) Break bird into parts. Thighs, breasts, tenderloins and “everything else” each into their own bag.
2) Butterfly the breasts and lay them over the skin. Rub with salt, butter and whatever herbs you like. Roll breasts and skin up into a roulade. Refrigerate for 2 days.
3) Sous vide breast roulade at 140 for 4-5 hours then deep fry.
4) Separately, braise the legs/thighs in red wine and mirepoix.
5) Use the bones/gizzards to make gravy.
6) Toss the tenderloins in the freezer for another day.
Ideas courtesy of Serious Eats. Works very well but does require a lot of upfront work. The results are worth it though.