Hacker Newsnew | past | comments | ask | show | jobs | submit | chrismustcode's commentslogin

They were failing as an online IDE for several years then growth shot up after the AI pivot.


Couldn’t they just send some hardware down Texas to co-locate there (presuming specialist hardware) and add another deployment target for their software? Would it be that hard?


The speed of light limits fibre speed which in turn limits high-frequency trading.

Flash Boys by Michael Lewis was a fun read on the subject. One memorable quote alleged that HFT traders would "sell their grandmothers for a microsecond [of edge]"


The issue is the speed of light.


for an interesting reversal of the "problem" of the speed of light, IEX is a stock exchange design to combat HFT by adding a physical speed bump by way of 38 miles of fiber optic cable. The general idea being to level the playing field and improve market liquidity using physical communication limits of light. https://en.wikipedia.org/wiki/IEX


That marketing gimic adds hundreds of microseconds to order latency. It’s not designed to level any playing fields it’s designed to get publicity.


Not really because anyone running a trading strategy that needs to worry about latency is already running their servers in the same datacenter as the exchange, so that just moves with it. What probably is an issue is that the datacenters required for a market don't look like AWS datacenters. I don't have any direct experience here, but I would be shocked if HFT software is something you could just deploy to a standard VM like on AWS.


They'd probably be running in an Equinix facility instead of AWS.


I thought they use GPU for learning and TPU for inference, I’m open to been corrected.


The first tpu they made was inference only. Everything since has been used for training. I think that means they weren't using it for training in 2015 but rather 2017 based on Wikipedia.


The first TPU they *announced" was for inference



no. for internal training most work is done on TPUs, which have been explicitly designed for high performance training.


I've heard its a mixture because they can't source enough in-house compute


I’ve never had an experience in any house or office where there’s been enough sockets to leave everything plugged.


I've never had an experience in any house or office where anything has ever been unplugged other than to put it away (a kitchen appliance that doesn't need to live on a counter, or a hair dryer, for example).

Buy a fused extension cord with more plugs, you have now turned one socket into 4, 6, or 8 sockets. You can even get some that have USB built-in, so you don't use a socket up for a phone or tablet charger. They're not even very expensive.

And in an office, I'm pretty sure all equipment (computers, lights, controls for adjustable desks if you have them), are meant to remain permanently plugged in anyway in a properly installed desk setup. What is going on in your office where you're choosing what is plugged in and what isn't, constantly? And why can't your office manager spring £20 for an extension cord with multiple sockets?


I've never stepped on a plug myself, so I agree it's not a major problem.

However, some older houses in the UK have far fewer sockets than more modern properties - sometimes only one or two per room.

And sure, if you need to use a hairdryer and a hair straightener a person with an orderly lifestyle might return them both to a cupboard afterwards - but some people don't mind clutter and just leave them wherever.

When it comes to multiway extension leads - people in the UK are sometimes told it's bad to "overload" sockets but have only a vague understanding of what that means, so some people are reluctant to use them.


"When it comes to multiway extension leads - people in the UK are sometimes told it's bad to "overload" sockets but have only a vague understanding of what that means, so some people are reluctant to use them."

To be fair, most people work on the assumption that if the consumer unit doesn't complain, then it is fair game. They are relying on modern standards, which nowadays is quite reasonable. I suppose it is good that we can nowadays rely on standards.

However, I have lived in a couple of houses with fuse wire boards, one of which the previous occupants put in a nail for a circuit that kept burning out.

Good practice is to put a low rated fuse - eg 5A (red) into extension leads for most devices. A tuppence part is easy and cheap to replace but if a few devices not involved with room heating/cooling blow a 5A fuse, you need to investigate. A hair dryer, for example, should not blow a 5A fuse.


Hair dryer and straightener would both be on a counter, right? No stepping issue there. And the same for appliance switching.

The only thing I plug in at ground level that isn't semi-permanent is a vacuum. No plugs are left lying around all day.


I honestly find in go it’s easier and less code to just write whatever feature you’re trying to implement than use a package a lot of the time.

Compared to typescript where it’s a package + code to use said package which always was more loc than anything comparative I have done in golang.


A lot of these have business logic literally in the database built up over years.

It’s a mammoth task for them to migrate


Oracle Consulting gladly built it all as stored procs with a UI.


> built

billed


annually


I agree it should be across the spectrum where people have the same rights to privacy.

> those who can afford to make the government look the other way and those that can't.

> Congratulations to Apple on lobbying for its own money. Very noble.

But what’s your implication here, that Apple shouldn’t have fought it?


Probably that it should be a generalization and apple should have fought for that and not apply just to one particular operator.


As far as I know, the blue/green mentality is a cultural issue for Apple. They would be fine if Android users had their data read by the government, because that injustice is a market differentiator for them they can then sell.

I'm not saying they shouldn't lobby for what they believe in, but Apple always stops short of making the world a better place and seems to care only if their walled garden is secure.


> Apple always stops short of making the world a better place and seems to care only if their walled garden is secure.

succinctly summed up why I dislike Apple (despite using their products). If you value privacy (against third parties), E2EE, and the tight device coupling then Apple is literally the only choice unless you have the time, knowledge and desire to piecemeal together your own solutions and that really sucks. I have permanent cognitive dissonance because I won't give up the small quality of life features Apple gives me, but I also don't have the time nor skill to replicate their whole ecosystem with Linux, GrapheneOS, writing BLE scripts for watch unlock, fussing with KDE connect for universal clipboard, hosting my own nextcloud instance, etc.

I wish there was another choice of mobile + accessories that was both privacy respecting and actively using open standards for the betterment of all, not just their own profits.


> I wish there was another choice of mobile + accessories that was both privacy respecting and actively using open standards for the betterment of all, not just their own profits.

That's the rub. If you look at Android handset financials, there's almost no money in making Android phones unless the company making them is Samsung, and only certain models sell. Where are all of these profits going to come from?

I wonder if you'd get farther with a USB SIM adapter under desktop Linux in that regard. I think you'd be hard pressed to end up where you want to in anything more portable than a laptop, since phones themselves are designed to be glorified containers for your mobile ad ID.

https://en.wikipedia.org/wiki/Advertising_ID


s/secure/profitable/


You hit the limits on this like immediately (it’s 1000 requests which each tool call is a request and even then I’m convinced pro isn’t 1000).

Also data retention Gemini has which you have to download a vs code extension to turn off for the CLI.

It isn’t an enterprise product it’s a way to get data for tool calling for training as far as I see it (as it currently stands).


I’d be stunned if a 270m model could code with any proficiency.

If you have an iPhone with the semi-annoying autocomplete that’s a 34m transformer.

Can’t imagine a model (even if it’s a good team behind it) to do coding with 8x the parameters of a next 3/4 word autocomplete.



That was previous iterations, 2.5 is 1 million context window

https://ai.google.dev/gemini-api/docs/models (context window is details under model variant section with + signs)

They were meant to crank 2.5 to 2 million at some point though, maybe waiting now till 3?


Maybe consuming the resources internally.


I mean the Harry Potter books are 2M tokens.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: