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

The ratio of managers:staff engineers has been decreasing, so sort of, yes.

Then they get fired for poor performance and you hire new fresh people.

Rinse, repeat.


Writing to disk for every write is required, otherwise you're not durable.

Sure it's faster to never write to disk, then you reboot and you've lost data.

/dev/null is a webscale database that is even faster!


There are a lot of use cases where you only truly need consistency, and durability can take a back seat. RocksDB for example does not fsync its WAL writes in the default configuration.

https://github.com/facebook/rocksdb/wiki/WAL-Performance#non...


If you can't at least guarantee write ordering you don't even have consistency.

Fsync is often used when the data doesn't truly need to be on disk, because there aren't very good write ordering APIs exposed, even if that's all you truly need.


read the whole article. WAL is the transaction log and the author tested correctness after a crash.

Well, the thing about reliability is that you can't really guarantee it by testing one particular scenario.

It seems to me that neither the old nor the new version of the code is really "durable" as I would understand the word. The old version made a write syscall per batch, but doesn't say it also did an fsync per batch. The new version writes data to an mmap'ed file, and calls fsync in the background.

So both versions are "durable" in the sense that written data is preserved even if the process gets killed, because it's in the OS page cache. But in both versions, a write can be completed before the data actually makes it to disk, so a power failure will lose acknowledged writes.


They tested SIGKILLing the process, they didn't test a power loss situation.

"Every batch of writes called file.Write on the write-ahead log"

You don't write to the WAL on a batch.

> the author tested correctness after a crash.

You mean the LLM?


It is both absolutely gorgeous and luxurious, yet still at less than $6 million, pretty modest for someone capable of giving away half a billion.

I respect someone with a good, proper library. So many luxury properties seem to miss that one.

[flagged]


That's a very naive, but common, viewpoint of wealth. "Worth" 1.3 billion does not mean "has 1.3 billion lying around in liquid cash". Net worth is tied up usually in many bank accounts across multiple banks, securities, real estate, trusts, etc. And that's all excluding capital tied up in corporations/orgs. Freeing up and giving away half of a billion dollar net worth is a difficult and time consuming thing, one which requires effort to do.

I guess that’s why more billionaires don’t give money away, it’s just too hard.

I promise they can afford to hire someone to do the difficult and time consuming thing here.

They certainly can. And the list of people who have become rich only to have it siphoned away by bean counters is at least as long as the people who are still rich.

Well when you say it like that, he’s basically just one of us!

No, he's not. Unless, of course, you've founded a company of comparable size.

gotta upgrade your sarcasm detector

It's even easier to give away money when you don't have it.

It's easy for some to give away money if they have it, but it's hard to accumulate that much money if you're one of those people.

Newmark maybe representing the occasional exception, most people who accumulate great wealth through their effort and ideas are afflicted with an addiction problem and have a hard time saying "this is enough for me". They become attached to how much more wealth they might be able to gather and recognize how the wealth they already hold plays a role in making the most of that.

Even if they somehow imagine themselves as philanthropists or minimalists, they tend to put off giving away too much too soon under the rationale that they'll be able to eventually share more if they hold out and use it more practically.

And if you don't have that kind of mindset, you're probably just not going to climb your way into the level of riches we're discussing here. You might still do quite well as far as most people are concerned, but that billionaire milestone is hard to catch without some propensity for wealth addiction.


> Even if they somehow imagine themselves as philanthropists or minimalists, they tend to put off giving away too much too soon under the rationale that they'll be able to eventually share more if they hold out and use it more practically.

What you're describing sounds exactly like Effective Altruism. The issue is that the expected value of an act that happens at an unspecified time in the future is zero.


It is empirically not easy.

I'd be okay with someone who has it giving it to poor old me who has not

t's also really easy to get greedy or ambitious/selfish.

You would not, tho. The people that think they need to get rich first before giving away won't ever give away. They always want more.

> it's easy to give away money if you have it.

You'd think that, and yet so few of them give anything significant away (unless you count political donations).

For every one Newmark or MacKenzie Scott who does give there are hundreds who only use charitable trusts as a tax haven, if at all.

So when one of them manages to avoid the extreme-wealth-to-meanness pipeline theorized by Paul Piff (et al) I'm happy to recognize the good they are doing in the world when so many of their peers are going so hard in the other direction.


> MacKenzie Scott

What exactly has been achieved with the billions MacKenzie Scott has given away? She seems to do very little vetting of the organizations she donates to. And upon a cursory inspection, they mostly seem like the kinds of organizations where most of the donor dollars go to white collar non-profit employees that produce little in the way of results, with very little going to needy people.

I feel like the billions might be better used to piggyback on government programs that have already identified need. For example, we could offer universal school lunch for an incremental $11 billion a year (on top of the $18 billion the government already spends). I bet a few of these “giving pledge” billionaires could just fund such a program in perpetuity.


Not really interested in arguing about whether or not she is giving her money away perfectly or to the best causes according to your (or my or anyone's) standards.

She gets credit from me for doing actual philanthropy instead of just cosplaying it for tax purposes, which has been the norm even for quite a few of the other "giving pledge" donors.

As far as bolstering government programs go, I'd love to see billionaires doing more in that area, but I'd also like to see us tax wealth properly to fund the government doing the government's job.


I find that the people complaining the most about billionaires, don't put much effort into help others.

You can always give away labor, in the form of volunteering.


Giving away money is stupid. With that much money I would rather fund companies that push the frontier on something. If it fails no biggie, if it succeeds it will probably end up employing people and giving them a source of income for life while simultaneously contributing to human progress.

You're saying charities like OpenAI that people have given money to haven't pushed the frontier on anything?

You'd rather vibecode a SaaS or start a dropshipping business because that is somehow pushing the frontier more than a charity like OpenAI people gave away money to?


Why wouldn’t you fund nonprofits and artists?

Artists don’t need money. More money is poured into the arts than at any time history.

Motion picture arts, literary arts, video game arts, graphic design arts,

and also at no time has it cost less to get an audience and find supporters. YouTube , TikTok, instagram, twitch, patreon, starsubscribe, gumroad, etc…


A lot of non-profits are just jobs programs for people who run in the same educational/social circles as the wealthy people. Not all non-profits obviously. But a shocking amount when you get close to it.

And for-profit companies are better than a quality non-profit…how?

Note my caveat: "Not all non-profits obviously."

I didn't say anything about a "quality non-profit." I said that a lot of non-profits are jobs programs for people who travel in the same social circles as wealthy people. Those non-profits often make little meaningful impact. Those non-profits are worse than for-profit companies because ineffectual for-profit companies at least go out of business. By contrast, fund-raising for non-profits generally isn't based on results, it's based on social networks. Donors get the same tax write-off whether the non-profit is effective or not. And they have social reasons for donating to non-profits run by people they or their spouse went to school with, grew up with, etc.


Because you get very little value from them. I haven’t seen anything paradigm shifting come out of non profit besides OpenAI which they then promptly turned into a for profit spinoff.

The win for something like OpenAI isn't getting a ton of customers to pay $10-100/mo.

It's getting businesses to pay $2k/mo or more per professional employee, like a lot of Anthropic customers.

Anthropic is ahead of them there, but that is how they win.


> Anthropic is ahead of them there, but that is how they win.

Isn't Anthropic currently killing that market though? I've been hearing about a lot of businesses pulling back after having experienced the reality.


Wouldn't the real story be to get government contracts? Those are more immune to public fickleness and market competition and usually have truly ludicrous margins.

If they're the only ones who ̶a̶r̶e̶ ̶w̶i̶l̶l̶i̶n̶g̶ ̶t̶o̶ ̶b̶e̶ ̶t̶h̶e̶ ̶e̶n̶g̶i̶n̶e̶ ̶f̶o̶r̶ ̶a̶u̶t̶o̶n̶o̶m̶o̶u̶s̶ ̶k̶i̶l̶l̶b̶o̶t̶s̶ can draw a reciprocation dingle-arm to reduce soinosoidal repleneration, then "I'm sure the government will buy it" [0]

[0] https://youtu.be/Ac7G7xOG2Ag?t=89


> would you want claude to be your boss? would you want claude to be your coworker?

I've had worse. Mostly much better, but I've had worse.


People are rightly worried about that, but is there any indication that it nullifies any privacy contracts around the data? Is it:

1) We know that legally privacy terms to data are still binding, and those worried about it are freaking out over nothing,

2) We know that those contracts are null and void, and there are no restrictions on what can be done with that data beyond blanket legal protections to such biological data, or

3) It's an open legal question

I don't understand the legal terms of something like this in bankruptcy, if the data are seen as being separated from the contractual obligations that acquired them.


Which way do you think that goes? Are the ones who "get it" the ones who are captivated or see them as incremental?

I guess all of them?

Some people "got" LLMs back in 2022, others needed it to evolve a bit.

It's not unlike computers. I started using them back in the 90s and absolutely nobody I knew was interested, while today everyone carries one in their pockets...


If I wanted to use two circuits without running extension cords, every place I've lived would mean getting electrical rewired.

If you're gonna get rewired you may as well install a 240v circuit, and some 120v 20a sockets while you're at it.


You can break the tabs on a 15A or 20A duplex receptacle to have (2) single 15A or 20A dedicated circuits on a single duplex receptacle.

It would require an additional run of 14/2G romex (12/2G for 20A) and a single-pole breaker, but allows you to skip cutting in an old work box to add a 2nd duplex receptacle.

You could possibly replace the existing 14/2G with 14/4G which has enough conductors for both circuits.


If you are going to do that, why not install a NEMA 5-20R receptacle, that has two independent circuits and is backwards compatible, as well as being rated for 20A per plug?

The receptacle is the easy part, running the new circuit is the hard part.

Or you know, install a new 240V receptacle.

If I have to:

1) Run wire

2) Get a bigger breaker box

3) To do it legally, hire an electrician and maybe get a permit

Replacing the receptacle is like, <1% of what's involved there.


Breaking the tabs on the existing receptacle prevents one from having to use a jab saw or multitool to cut a hole in the gypsum wallboard or plaster and install a cut-in/old work box to add a 2’d duplex receptacle: https://www.homedepot.com/c/ah/how-to-install-remodeling-box...

Any 15A or 20A duplex receptacle can have the tabs broken to get two separate 15A or 20A simplex receptacles, you don’t need a 5-20R for that, a 3-wire 5-15R works just fine.

Someone upthread mentioned 1.2kW load which a 15A receptacle handles just fine: .8*120*15=1,440W continuous. Bumping that up to 20A only gets you an additional 480W of continuous load: .8*120*20=1,920W. A continuous load is one that runs for 2 hours or longer, the overcurrent protection and wire must be upsized by 1.25x (or derated to 80%)

Most receptacles in homes are wired with 14/2 romex which is only good to 15A (in homes, which use the 60C ampacity column) which is why I suggested pulling another run of 14/2G romex and breaking the tabs. Pulling 14/2 romex to an existing receptacle usually isn’t that hard if you have a fish tape.

AFAIK computer PSUs can’t easily use 240V power without a PDU in the middle, but I’m likely wrong on that, especially for server PSUs.


Almost all computer PSUs I have ever seen are 110/220 since they don't make different models for Europe.

Gotcha, that would make sense. In that case, your suggestion of a 240V outlet is best. Swap the outlet for a 240V one, swap the breaker out for an 240V 2-pole, and use the same wire, assuming the wire is already big enough since you won’t need a neutral.

Fair point.

I'm very close to just running a cord over or devising a way to put my machine closer to a second circuit because my rental is horribly setup and both my bedroom AC and living room desktop (that also doubles as a ML training box) end up on the same circuit.


Population growth isn't limited except through resources but AI also needs resources

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

Search: