I mean that 'almost as fast as Groq' is not a compelling strategy. If it's almost as fast, why not just go with Groq? Are they going for cheaper? Added services? I'm just wondering.
Groq quantize their models, killing accuracy. SambaNova run the models at full precision, and on a single node, whilst Groq runs a single model on hundreds of chips. SambaNova run training and inference, Groq only run inference. SambaNova run hundreds on models on a single node without any performance degradation, Groq need hundreds of chips for a single model.
Groq are trying to launch their own cloud and running it at a loss to try to acquire users are raise venture capital so they can tape out a new chip (their current chip just is not very competitive, they get speed for having a good interconnect, but it is so cost ineffective). SambaNova sell systems to big companies and service providers, and don’t have a paid cloud API, so aren’t as visible as Groq. But SambaNova make real revenue whilst Groq have literally zero revenue, which is why they struggle to raise money.
Very soon people will realize that Groq are frauds and SambaNova is the legit challenger to Nvidia.
Well, Groq memory-per-rack looks very low. That is to say, the whole world now understands how exciting very high inference throughput is in a way that almost nobody did when Groq started, (I think I saw an early pitch deck, and don't recall fast inference as even a differentiator in that deck, although I could be wrong). However, the number of Groq chips/servers/whatever you call them that are needed to get up to running Llama 400B looks like a lot. Like many, many racks worth.
Plus, Groq claims to have converted over to only being a cloud provider now, and will keep their hardware to themself. Given that I can't even sign up for a pay-as-you-go API key with Groq right now, I think there's a lot of room for competitors.
This is just a sad consequence of some merchants creating accounts in the thousands/hundreds of thousands and selling them to SEO/Marketing blackhats to use.
Google should simply have paid, in-person support for these issues. Cuts off the spam merchants, while still letting Joe Average create an account with their usual phone number once they've been verified as an actual person. That's got to have some value, even to Google themselves.
I don't think that would go over well. When Google asked people to share a national ID or credit card in the EU (age verification is legally required to be compliant with the new "think of the children" regulations in the EU's Audiovisual Media Services Directive), people online freaked out.
The number of simple, yet very cool ideas on HN just baffles me, no wonder HN is an addiction!
Selling an add-on/accessory that clips onto any macbook or even any laptop with similar screen properties as the macbook pro used will be a nice little business!
> MirrorMeThis is a specially-designed laptop mirror. Instead of your students seeing your face, they see the keyboard area where you can place anything you want to share and talk about.
So it's designed for the classroom, but I don't see why it couldn't work here.
Since it is this time of the year again, here is an idea: try bouncing the view of the screen off of a metallized glass Christmas bauble for a makeshift curved mirror.
Edit: an iPhone fails to focus on the reflections in a Christmas bauble, maybe the warped reflection throws something off in the auto-focus method.
Edit 2: it works with a larger bauble, supposedly because the view is not as warped as in a small one that has a smaller radius of curvature.
I remember 2008 Prius had mirror to reflect Speedometer. Basically Speedometer is on the floor of the dashboard and a vertical mirror reflects the numbers. I assumed they made it this way to save space.
This technique is often used to reduce eyestrain and refocusing by extending the optical distance to the display, making it.more similar in terms of vergence angle and focus distance to the view outside. I'm not sure if this was the motivation in the Prius but given the fairly advanced state of VFD displays at the time I'd guess so, there would have been options to avoid the complexity. It's further curious that the Prius used to use half mirrors to combine multiple display units for that dashboard, I'm less sure of the reason for that other than that perhaps the assembly was expensive and they saved money by doing it only for critical instruments like the speedometer.
Correct me if I am wrong, but is there anything stopping anyone from starting a new (better) search engine, and promoting it like crazy? Maybe even pay users a portion of the ads revenue for using the new search engine?
I understand there are economies of scale here, but if a few investors got together what's to stop them from achieving this?
Nobody is stopping anyone from doing this, and people are doing this. Whenever I lock my screen on Windows, I get an unremovable ad from Microsoft that says I can earn money for every search if I just switch to Bing. When you click "help" in Windows, it doesn't give you help -- it opens your a browser for a Bing search for something vaguely related to the problem, and the search returns no useful results. (It doesn't link to as much malware as it used to. Maybe that's because of my ad blocker though.)
The reality is that Bing sucks. Hard. It can't even return Microsoft's knowledge base for a search term that Microsoft explicitly set up. Every time, I cut-n-paste the term into Google and that is how I resolve my question. (Of course being linked to Microsoft's own website! They wrote the software. They wrote the documentation. But their own search can't find it!)
The problem is not Google being anti-competitive here. It's Microsoft that can't compete. They don't have two employees that said "hey, we should set up a monthly sync to make sure all the help in Windows gets linked to our knowledge base in Bing" between someone on the Windows team and someone on the Bing team. And that's the kind of thing that makes your company go out of business. The government can do very little to help you with that.
Ah, I see that now. It came back after my reinstall. (I'm good at turning off Cortana and sending my web browsing history to Microsoft. This one I missed, however.)
People use Google because it's what they know and the alternatives for whatever reason do not sufficiently differentiate themselves in ways that the user cares about enough to switch.
In order to be disrupt Google, you don't make a better Google. You identify something Google doesn't deliver and which people care about enough to use your search engine. I don't think anyone knows right know what that hypothetical killer feature is, and most probably assume it doesn't exist.
On another level, though, incumbents have a habit of buying potential threats (which can be a pretty good deal when you're in venture capital land and need to show your shareholders high returns). Google's bought a number of companies with a different take on search over the years, such as Like.com and Clever Sense.
The reason Google is because of their AI. And the reason their AI is good is because of the massive amount of data they collect by violating privacy. No other company can start right now. Their search engine is years of efforts of collecting every user's data in Chrome for determining 'freshness', page rank and other related stuff.
You are making the same flawed argument that the DOJ complaint is making: that Microsoft, Apple, Facebook, and Amazon, some of the largest companies Earth has ever seen, somehow can't afford to compete in search, which is plainly ridiculous.
The DOJ complaint isn't "You can't have a monopoly". In fact there is nothing illegal about having a monopoly in a product or service under US law.
The problem arises when a company leverages a monopoly to suppress competition, implement predatory pricing, or manipulate competition in another market. Using a monopoly in search to compel tying agreements prohibiting browsers or manufacturers from including other search engines or apps would qualify. If it turns out that agents of the company (Directors, VPs, etc) internally communicated an intent to suppress competition that would be more than enough to bury them.
They could compete, but it would be a money hole. You’d have to spend billions and years to reach some level of parity with Google to persuade enough people to switch, and even then the marketing spend necessary to reach any significant level of switching would be another massive investment.
"wah, this is hard" is not an antitrust case. Anyway it's clearly not all that hard since Bing exists. The fact that Apple sells iOS users' attention for six billion dollars a year is strong evidence of Bing's strength. Market prices are set by the 2nd bidder, not the highest bidder. The price Google pays to Apple is what those eyeballs would have been worth to Microsoft, not how much they are worth to Google.
The quality of google search is good for consumers, it also isn't anti-competitve to improve the quality of your product, even if you're using resources and knowledge from unrelated businesses to do so.
The ideal outcome of this lawsuit is Google being forced to open up their vast amounts of data obtained by being the 90+% market monopoly.
Only then can other search engines even begin to compete.
Theoretically there’s nothing stopping them. But I personally think realistically it’s too far gone for this current generation. Computer literacy is too low.
The same way an entire class of people (looking at you, Grandma) think that Microsoft Windows IS the computer, many people think Google IS the internet.
And to some degree they aren’t wrong.
Overcoming that perception is going to be insanely expensive and time-consuming. Not to say it shouldn’t be done. Not to say it couldn’t be done, but man, talk about an uphill battle. Especially as Google will naturally mobilize their own machine to crush this new competition.
I guess the question is, to what degree is this like worrying that people identify Xerox with photocopiers? People are still gonna be saying "googling" in a decade, certainly, but there are lots of ways that could go that don't help Google as a company.
I think the biggest difference is the degree of interaction. I'm 25 and have used a photocopier a whopping 5-10 times in my life. It's a non-impact.
The internet, however, is different in both the frequency of interaction and the import of being transparent.
Also, as far as I can tell, it would be an order of magnitude cheaper to start a new photocopying machine because the tricks and tactics that Xerox could ostensibly use to slow down your new startup are relatively benign.
Google controls (more or less) the way anyone, anywhere (who isn't tech-savvy) will find you as an online business. Kinda scary.
> is there anything stopping anyone from starting a new (better) search engine
Nothing, but that's always been the case, with every monopoly. There was nothing stopping anybody from running telephone lines and competing with AT&T in the 80's except for AT&T's massive incumbency head-start.
You chose the perfect example to disprove your own point. Other companies could not run phone lines. There are plenty of hard monopolies, and they’re very different from soft, virtuous cycle monopolies.
Hm, ok, you're right - there was regulatory capture in place at that time, but when the phone companies started running lines in the first place in the 1800's, they had to work around the local regulations at the time as well. Worth noting that the AT&T breakup didn't involve restricting the artificial monopoly, but limiting the power of a single entity to control the market as they had.
As described in the article: It's a chicken and egg problem. More people using a search engine, leads to more data for analytics, more analytics lead to better results and again more people using it.
The part you chose to put in parentheses is the hardest part. No matter what you did to improve the search experience, Google could match it at lightning speed because of their scale.
The argument is that Google would use its monopoly in other areas to defend its monopoly in search. For instance, what if the most popular browser (Chrome, conveniently for Google) just... didn't work very well with this new search engine? Entirely innocent technical details, of course... a wonky library here, a faster protocol there which they can adopt first. Pretty soon, the new search engine is "the slow one" through no fault of its own.
yeah I imagine once a technique is mastered, it would be used all over the place! Not sure if one "company" started offering this service in many places, could explain the similarity.
These projects employed skilled journey men (level between apprentice and master). At some point they then traveled to work on the next project, spreading the know-how. Once something was working, it was often copied as modifications were risky.
To add to this: those were highly sought after specialists with a lot of privileges. This had consequences that sound strange to us nowadays.
Stonemasons were heavily involved in medieval diplomacy, since everybody in power needed them to build or repair fortresses. They could travel freely between kingdoms that were at war with each other and exchange messages between the respective rulers.
I'd rather call the knowledgeable entity in question "expert" than company, but I was wondering along similar lines: would they travel far to projects or would the knowledge be dispersed to enough people that most projects could just happen with local expertise? (plus temporarily imported capacity for big project, but not necessarily imported expertise?)
What's clear is that knowledge transfer (from region to region and from generation to generation) was facilitated by traveling apprentices, a tradition that still lives on and (I believe) is documented to go back to that time. Their profession wouldn't be bridge builder but merely some contributing role (with the occasional exception of whatever the period-correct term for architect or project manager would be), but of all the masons involved, you'd have some whose learning circuit involved a bridge, of the woodworkers you'd have some who dabbled in cranes before and so on. But was that usually sufficient or was there also a pattern of on-demand traveling leadership?
edit: post was lingering in the input form for some hours, some good answers already there!
That does not necessarily mean only a small set of neurons in our brains are engaged.
It could be that we use the whole set or a large portion, albeit in a more efficient way!