What are your primary usecases? Are you mostly using it as a chatbot?
I find gemini excels in multimodal areas over chatgpt and anthropic. For example, "identify and classify this image with meta data" or "ocr this document and output a similar structure in markdown"
I run a bunch of smaller models on a 12gb vram 3060 and it's quite good. For larger open models ill use open router. I'm looking into on-
demand instances with cloud/vps providers, but haven't explored the space too much.
I feel like private cloud instances that run on demand is still in the spirit of consumer hobbyist. It's not as good as having it all local, but the bootstrapping cost plus electricity to run seems prohibitive.
I'm really interested to see if there's a space for consumer TPUs that satisfy usecases like this.
I also noticed this as well. Actually went back and watched it several times. It's an incredible moment. I keep thinking, "if this moment is real, this is truly a special person."
The gains may be low, but they do still hold most of their value. So it's one way to at least "lock in" big gains from the last few years, and also to "wait out" and survive a major stock market drop so you can "buy the dip."
> I don't understand why people love the nanny state so much. We can't continue to make companies be the police, the stewards of truth, and justice. They demonstrated just recently, during COVID, that this was an absolute disaster. Over the last 30 years we have watched freedom erode because the average American wants to foist all responsibility onto someone else.
I think there is a couple of things at play:
First, negativity bias. I think it's pretty clear that as a society we're not that interested in harm reduction, just biased towards harm reduction of things that violate our value system. So when things happen that do violate our norms, they're presented outside of the background noise. For example, very few people feel compelled to come in and share personal anecdotes of how they lost relatives to a car accident when the topic at hand is vehicles in america. Yet they're the second leading cause of death from unintentional injuries.
Second, these things affect people across the social stratification index. People of privilege experience it. I claim that we also as a society are not very concerned with protecting vulnerable populations. The top 10% of the nations families hold 60% of the wealth, while 1 in 10 Americans live in poverty. We consistently rank lower in social welfare compared to other developed nations. So, further supporting the first point, it's even more outlandish when these things happen to people who are not accustomed to having bad things happen.
Finally, technology consistently outpaces our ability to reason about and structure our society as a whole. It's easier to attempt blanket and ham fisted reactions to these bad things we see without understand the wider implications.
To a lot of people, the easiest and most obvious choice is Authoritarianism, because in there mind there's no other way to stop the pain.
Plus, it's difficult to talk about these things without being callous. "Bad things happened to me, so you should simply give up your right to privacy so we can prevent it from happening further." At face value is difficult to take seriously, but when it involves that cross section of the privileged vulnerable class, it's difficult to have a reasonable argument without being steamrolled.
I use it too but 106 "e.g."s in a single page? That's how many there are now. Not to mention it's full of inconsistencies and being edited multiple times.
I think the author might have left an LLM agent in a loop fixing it whenever HN points out an error or finds something new to add on the internet.
As someone who has been seriously depressed from an early age, I can tell you that it looks exactly like the DSM/ICD criteria - a lack of energy, loss of appetite, loss of interest in all activities, insomnia, feelings of worthlessness, suicidal thoughts and pervasive sadness and hopelessness.
Some people would rather believe that pediatric depression isn't real, rather than confront the reality of a loved and cared-for child who is constantly tearful, severely underweight, sleeps for three or four hours a night, spends most of their time staring into space and frequently talks about wanting to die.
Depression is an utterly dreadful illness and should not be confused with normal sadness or unhappiness.
Probably something like Boy Interrupted[0]. Sad story and something I can sympathize with having some of the same feelings very early on despite having a rather normal upbringing and siblings not showing signs of it.
And several of your comments before that were upvoted. Are we to regard those as suspicious?
Of the recent downvoted comments, one was a complaint about moderation that anyone who has paid attention to dang's track record here over more than a decade knows is baseless. (And if you think the top comment on any thread is a bad one, you can always choose to be a helpful contributor to the community and email us to let us know).
Of the other two of your downvoted comments, none were downvoted by the same users.
The choice is yours to make an effort to observe the guidelines and be a positive contributor to HN, or alternatively to keep using HN for political/ideological battle and complain to the moderators when things don't go your way, but it's clear what others in the community want to see.
> or alternatively to keep using HN for political/ideological battle
Which ones? The one about ML and programming languages? Or the one about asking a genuine question about an experience with childhood depression? Or the one observing that you and dang unevenly apply moderation rules? Or the one commenting about how you can't say the word for the literal definition of fascism on this site without getting downvoted? Or the one about dishwashers?
Where's my ideological battle?
You have no credibility. You unlike dang, don't do a good job. Go ahead and ban me or put me on a cool down to prove my point.
I've scanned your full list of comments and can find plenty that have an ideological flavor to them, and others that are in the flamewar style, but are not so clearly related to politics/ideology. I'm not interested in getting into an argument about which of your comments are ideological or not. That's not the issue. What is the issue is the hostile and inflammatory style of commenting towards other community members and HN as a whole.
It's notable in this instance:
- You posted a series of comments about controversial topics, having established a history of participating on HN with this persona of being a brave combatant for, I don't know, some worldview or philosophy that you seem to be fighting for;
- When a handful of your comments receive even a solitary downvote, you call in "the cops" (dang) to come to your aid, with a claim of "brigading";
- When we investigated and found that, no, there's no "brigading", some of those comments are not even net-downvoted anymore, and that any downvotes you're getting are to be expected given your combative style of commenting, you've responded with these incoherent attacks on moderation/moderators.
Whether we all agree that many of your flamewar-style comments really are, in fact, political/ideological, is not the point and seems to be a way for you to deflect from being held to account for your conduct.
What I'm saying to you is that people who care about making HN better have all kinds of ways of showing it, and it begins with making an effort to observe the guidelines, and it also involves engaging respectfully with other community members and the moderation system. We are always, always working to make HN better and our moderation approaches better, and we always welcome and engage with feedback, as dang has been doing with you in another subthread today. But we've both been doing this job long enough to sense when someone isn't really wanting to help make HN better at all.
> What is the issue is the hostile and inflammatory style of commenting towards other community members and HN as a whole.
Please.
> having established a history of participating on HN with this persona of being a brave combatant for, I don't know, some worldview or philosophy that you seem to be fighting for;
What? Just because I have a different worldview than you, doesn't mean I am fighting for or am a brave combatant of anything. But it's extremely telling that you think that, and revealing about your own views. And furthermore troubling that you are a moderator here.
Maybe you should read up on the clustering phenomena wiki and understand your own personal biases a little more.
You don't know what my worldview is or what dang's worldview is and honestly I don't know what your worldview is and this is never relevant to how we moderate HN. We want HN to be a place where difficult topics can be discussed and all perspectives can be represented. That's what we optimise HN for, with the caveat that the guidelines foremost ask us all to "be kind" in comments. It's notable that you keep complaining about some kind of "bias" without being able to point to any evidence for your claims, and that all of your comments in this subthread ultimately resort to ad hominem. If there was any substance to your claims you would have presented it by now. The entire history of HN submissions and comments is available for anyone to download and analyse.
Let's be clear what's going on here: you've claimed to be a victim; you can't demonstrate exactly how you've been made a victim; when we investigate your claims, which we've taken time to do in good faith, we find that, no, there's no evidence for your claims of victimhood; when we tell you that, you respond with ad hominem attacks.
Please just observe the guidelines like everyone here is expected to do.
I've been here on this account for five years. Making me out to be some kind of serial complainer and self-proclaimed victim, of which I've done exactly twice across a litany of diverse and continued conversation and dialog on this website is ludicrous.
Your continued aggressive dismissal of milquetoast commentary against your moderation style is offensive.
Your characterization of my posts here as a warrior championing some cause is similarly offensive.
Your words, not mine:
> ... having established a history of participating on HN with this persona of being a brave combatant ...
How kind, and full of good faith.
You clearly feel something towards my worldview. Your language is charged, you have opinions directed at me. To be clear: I don't care about you at all. But I do find it amusing to watch such a visceral reaction to a general commentary, "the mods are biased, and shape the bias of this website."
You initiated all this. We've investigated and considered all your claims, established them to be unfounded or baseless, and still you keep going.
Your worldview is irrelevant to this discussion. This place’s entire worth is built on the fact that a broad range of worldviews and discussion styles is represented, and our moderation philosophy is intended to allow everyone’s worldview to be fairly represented.
We're responding to your claims only because if there is any basis to them we want to know so we can address any issues and reform the way we operate. We’re actively working to do that continually. But it’s increasingly clear that none of your claims hold up to any scrutiny and every additional comment just generates more noise and still no signal.
My advice as a long time participant here: pay no attention to upvotes or downvotes. Sometimes they seem to be completely unrelated to whatever you said. Stay curious.
It's always curious how and when you decide to pop into threads and request that people follow the rules of hackernews. You claim that the site and your moderation principles are not (or have limited) ideologically motivated(tions), but your enforcement (or engagement) is uneven and certainly along some political axis.
From the rules:
> Please don't complain that a submission is inappropriate. If a story is spam or off-topic, flag it. Don't feed egregious comments by replying; flag them instead. If you flag, please don't also comment that you did.
Yet one of the top comments of most front page items is always a useless comment of clickbait or some pedantic complaint/accusation about some format of the title/submission.
You have a hard job, it's not intended to be an indictment of your behavior. Just a general observation that I wonder if you're cognizant of.
If the community needs this so badly, why is the above aforementioned behavior so prevent that it's become a meme of hackernews behavior?
People are constantly (over)interpreting this "mods are against my side" bias into what we do (https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...), but I can tell you for sure without knowing which side you're on (or what the topic even is) that people you disagree with feel just as strongly that we're clearly/obviously/certainly against them and that we tilt the field towards you. It is by far the most consistent phenomenon I've observed in years of doing this job.
You're right, of course, that moderation is uneven, but the chief reason for that is that we don't come close to seeing everything. Beyond that, we no doubt have our biases (though different mods have different ones), but we also work hard at suspending them when moderating and have many years of practice at doing so. Many of our "you've unfortunately been breaking the site guidelines repeatedly lately" replies are addressed to commenters whose position on an underlying topic we actually agree with.
I would never claim that we are perfect at being even-handed—this is impossible—but it's nothing as crude as what you think you're seeing here. That explains why the people you're most implacably opposed to also believe they're seeing the same thing, just in the opposite direction.
It's even more curious that merely pointing out that you unevenly moderate triggers yet another link to the clustering illusion.
The truth is: we never will be able to know the biases of this community or mod team with accuracy, because hackernews doesn't expose enough data to be able to perform a meaningful analysis.
The call-out is disjoint from the rest of what you go on to say. I said you are _certainly_ biased. I didn't say how. Your comment starts with saying that people are predisposed to feel persecuted due to biases, which is somehow related to my use of the word, and then you go on to essentially confirm exactly the intention of my callout.
> That explains why the people you're most implacably opposed to also believe they're seeing the same thing, just in the opposite direction.
Cordially, I have no idea what you're talking about or referring to in this regard. Who is my enemy that you've invented here? I'm not representing a position on behalf of the community, other than to point out that this moderation has _some_ uneven biases, and it's always interesting when they show.
In the words of @tomhow
> The choice is yours to make an effort to observe the guidelines and be a positive contributor to HN, or alternatively to keep using HN for political/ideological battle
What's my political and ideological battle? Functional programming? Ai usage?
When political topics come up, I engage with the discourse. That's within the bounds of the spirit of this community.
An observation of when the job is selectively exercised. It's clearly not evenly applied across all of the rules. There are some obvious applications, and there are some non-obvious applications.
As a libertarian, I'll stop calling them fascists when they stop calling themselves conservatives and actually adopt some kind of honest label for what they stand for. But that would require them to stand for something constructive rather than simping for whatever destructive looting Dear Leader has divined this week.
I find gemini excels in multimodal areas over chatgpt and anthropic. For example, "identify and classify this image with meta data" or "ocr this document and output a similar structure in markdown"
reply