riku_iki

32 ms·

691.
by riku_iki 2y ago
Gmail is super important source of ads targeting data. They won't kill it.
692.
by riku_iki 2y ago
This won't prevent them back to paywall in future if investors ask.
693.
by riku_iki 2y ago
There are some contenders in that list: TiDB, YugabyteDB, YDB.
694.
by riku_iki 2y ago
> Go is good for almost everything. Java is good in more areas than go
695.
by riku_iki 2y ago
> I don't doubt the results but I cannot imagine how that can be true, I'm not fit but I've had many 20+hours hikes to the high mountain peaks but if I run to a bus stop you are probably not trained as well as our ancestor
696.
by riku_iki 2y ago
> I am curious, do you think there is an ethical way to be a middleman like this? I think ethics is not applicable here. Corps are for profit enterprises, and seek every chance to bring gain to shareholders. Regulators goal is to support
697.
by riku_iki 2y ago
I am wondering why Apple and Patreon even a thing in this market. Most of the content is distributed through Youtube, so Google can just step in, create patreon interface and cut both of them, get some extra revenue, and be a good guy.
698.
by riku_iki 2y ago
> Another point that Patreon isn't really emphasizing here that seems relevant to any conversations about "fairness" is that Apple's fees on Patreon subscriptions in-app are now higher than Patreon's fees. they a
699.
by riku_iki 2y ago
For-profit corp needs some incentives for such donation.
700.
by riku_iki 2y ago
> search engine the default option on smartphone web browsers violate US antitrust law Interesting that they limit the case to mobile browsers. I guess MS share is significant enough on desktop.
701.
by riku_iki 2y ago
> Apple will lose significant revenue from having Google as the default search engine and Google will no longer have to pay those billions. there is also Android ecosystem, where DOJ may ask google to prompt user with Search Egnine selec
702.
by riku_iki 2y ago
I think everything was running in the same address space in those CPUs, thus no separate processes. But I am not an expert.
703.
by riku_iki 2y ago
info becomes public when it is officially announced by some of the transaction participants or disclosed in some fillings. I may be too skeptical, but it is just hard to believe someone (who?) throws 400M (to seed + A investors) plus signif
704.
by riku_iki 2y ago
I don't trust that site. They spread unconfirmable rumors behind paywall. Base on those numbers, $150M of series A with 1B valuation would be bought now for 375M(which is high for failed startup), my understanding is that such transact
705.
by riku_iki 2y ago
I think he just steps down as CEO, there is no company acquisition happening, so no one gets paid. I think company essentially failed to generate revenue and/or get new investment round.
706.
by riku_iki 2y ago
> I'm puzzled. Did Noam get $500mm+ from this? where did you get $500mm number?..
707.
by riku_iki 2y ago
> In particular, GPT-2 to GPT-4 spans an increase from 'well read toddler' to 'average high school student' GPT-2 was indeed much smaller and weaker model. But the question do we have "exponential" boost aft
708.
by riku_iki 2y ago
> The National Lawyers Guild was also an observer they sent 5 observers only per https://truthout.org/articles/maduro-wins-reelection-for-thi...
709.
by riku_iki 2y ago
AGI may find much simpler, more robust/performant and safe language.
710.
by riku_iki 2y ago
It can be accurately measured if attribution is helpful or not. You create two Ad campaigns with two different landing pages, one with attribution optimization and another is without, do A/B test and see which one will bring more conve
711.
by riku_iki 2y ago
> is there a specific reason you're not using plain-vanilla filesystems which _are_ stable? my personal reasons are raid + compression
712.
by riku_iki 2y ago
> Development taking long usually means that the model itself is too complicated to be done right in a reasonable time yeah, features rich/complete fs is complicated, that's why we have very few of them.
713.
by riku_iki 2y ago
but they likely put money into features which they are interested in, and not raid56
714.
by riku_iki 2y ago
> It's funny because Facebook uses btrfs for their systems & doesn't have these issues. they likely have distributed layer on top which takes care of data corruption and losses happening on specific server
715.
by riku_iki 2y ago
and we don't have 100% accuracy in translation in ambiguous texts, because system often need some domain knowledge, context etc. And math has 0% tolerance to mistakes. I also expect that math formalized by machine will be readable by m
716.
by riku_iki 2y ago
> Formalization is in principle just a translation process and should be a much simpler problem than the actual IMO problem maybe not, because you need to connect many complicated topics/terms/definitions together, and you don&
717.
by riku_iki 2y ago
> They are building a general pipeline which turns informal natural lamguage mathematics but this part currently sucks, because they didn't trust it and formalized problems manually.
718.
by riku_iki 2y ago
its not clear if theorem is actual input formal definition, or formal definition was in different form.
719.
by riku_iki 2y ago
Example of proof from AlphaProof system: https://storage.googleapis.com/deepmind-media/DeepMind.com/B...
720.
by riku_iki 2y ago
> but reddit has the non-exclusive right to redistribute and license the content. that's what I said: it is legal.

More ›