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

I think the only thing it's missing is the board view.

The way these are pushed as "solutions to cities and traffic" make making fun of the too easy.

> But it misses the social context

Funny how their entire social context is "never encounter another human as you go from A to B"


> way these are pushed as "solutions to cities and traffic" make making fun of the too easy

It's funny. It's also dumb. An observation can be both at the same time--it's a cornerstone of humor. What it isn't is fundamentally true or revealing.

> their entire social context is "never encounter another human as you go from A to B"

Nope. It's recognising that humans have diverse and varying needs for interaction and privacy.

I like to dine out, even alone. That doesn't make everyone who eats at home alone an idiot. (That doesn't mean I can't make jokes about it. But they shouldn't be mistaken for truth.)


> What it isn't is fundamentally true or revealing.

Well, they are not a solution to transport problems, or to traffic jams.

Yes, they can be complementary to other types of transportation. Yes, companies will enshittify them beyond measure if/when they reach a certain proportion of cars.

> It's recognising that humans have diverse and varying needs for interaction and privacy.

No. I don't think this was even uttered by any of these companies.

Waymo claims to be committed to safety: https://waymo.com/about/

Tesla: stress and safety https://www.tesla.com/fsd

Zoox: purpose-built taxi shaping the future of transportation https://zoox.com/about


> they are not a solution to transport problems, or to traffic jams

Nor to world hunger.

> companies will enshittify them beyond measure

A hypothetical applicable to every mode of transit, private and public.

> don't think this was even uttered by any of these companies

Things can be true without being in a corporate press release. (Also, you're the one who originally argued these services' "entire social context is 'never encounter another human as you go from A to B'." If not being in a press release is an argument against one, it 's an argument against the other.)

Though, in this case, it has been said: "Waymo gives you your own personal space to focus on more meaningful things" [1].

[1] https://waymo.com/rides/


> Nor to world hunger.

Ah, the good old ad absurdum.

These companies literally hail themselves as "future of transportation".

> A hypothetical applicable to every mode of transit, private and public.

These are private companies looking for profit. These are not hypotheticals given what is happening to other cars and car manufacturers.

> Also, you're the one who originally argued these services' "entire social context is 'never encounter another human as you go from A to B'."

These are literally robo taxis. A taxi is literally a car that is taking you from A to B. And they are also removing the driver from them. Oh, and don't forget the existing of things like Boring Co. which exists almost solely to undermine public transport.

Their intended future is nothing but endless roads with isolated vehicles going from A to B. There's no other "social context".


Don't show the discussion to John Carmack. He's baffled why people are so allergic to debuggers: https://youtu.be/tzr7hRXcwkw?si=beXGdoePRkbgfTtL

I'm pretty sure in that interview at some point he realized becasue the debugger experience for developers using Linux sucks compared to Windows where he does most of his work.

Alot of programmers work in a Linux environment.

It seems like windows, ide and languages are all pretty nicely integrated together?


> It seems like windows, ide and languages are all pretty nicely integrated together?

Not only, and not really. After all, for all its warts Visual Studio is still a decent debugger for C/C++. IntelliJ has pretty good debuggers across all of their IDEs for almost all languages (including things like automatically downloading and/or decompiling sources when you step into external libraries).

Even browsers ship with built-in debuggers (and Chrome's is really good). I still see a lot of people (including my colleagues) often spend inordinate amounts of time console.log'ing when just stepping though the program would suffice.

I think it's the question of culture: people are so used to using subpar tools, they can't even imagine what a good one may look like. And these tools constantly evolve. Here's RAD Debugger by Ryan Fleury: https://threadreaderapp.com/thread/1920345634026238106.html


I keep seeing people say this: "I pay for YouTube Premium. For my money, it’s the best bang-for-the-buck subscription service on the market" and I don't understand.

For me, Premium's only value proposition is removing ads. Recommendations are still the same (quite shitty). Search is unusable (4 relevant results then unrelated recommendations). Shorts are pushed aggressively no matter how many times you hide them. Search in history will often not find even something you just watched a few days ago.

It's the same Youtube.


Yes basically all it does is remove ads. Those of us who are happy with it are those of us who don’t feel entitled to unlimited video streaming for free. Most people think YouTube should just be free and have no ads for some reason, and they probably wouldn’t say Premium is such a great deal.

The issue is you're still not paying for the content nor paying the creators.

You're paying YouTube to stop annoying you, and they then decide what to do with that money, incidentally paying some creators.


Premium pays out to creators by minutes viewed (vs AdSense which pays out by ads viewed)

I've heard some creators say that in total, they make more money from all their Premium viewers than they make from all their AdSense viewers, even though the former are a small fraction of the latter.


This argument is repeated on other comments as well, but I think it's fundamentally a parralel fact.

YouTube giving some of the Premium money to creators doesn't make Premium a good product. If'm not that utilitarian to think any single additional penny going to some creators is good whatever YouTube takes in the process and the general impact on the the whole field.


YouTube seems to be pretty explicit that it is paying 55% of revenue from watching videos to creators:

> If a partner turns on Watch Page Ads by reviewing and accepting the Watch Page Monetization Module, YouTube will pay them 55% of net revenues from ads displayed or streamed on their public videos on their content Watch Page. This revenue share rate also applies when their public videos are streamed within the YouTube Video Player on other websites or applications.

https://support.google.com/youtube/answer/72902?hl=en#zippy=...


As you point out, that revenue split has a set of conditions, which also require a level of contract on Youtube and other requirements (not being DMCA stroke for instance)

So where does your Premium money go when you watch a very small creator ? where does it go for a demonetized video ? etc.

That might sounds like a subtle difference, but consider the gap with channel membership, super chats (which are also roughly 50% split I think?) or patreon for instance.


> where does it go for a demonetized video

A "demonetized" video is technically called a "limited or no ads" video in YouTube Studio - it means YouTube has determined that advertisers do not want their ads seen on the video for reputational reasons. Premium views still pay out for them since they are not paid through showing ads.

A DMCA strike is something else.


Sorry I wasn't referring to videos the creator decides to forgo revenue, as you point there's a better term for that.

I was thinking about the videos that were supposed to make money but got shut off monetization for whatever reason. DMCA strike is one, YouTube flagging it as risque is another common one.


I was referring to your latter example - when YouTube decides a video is inconvenient, it means they're afraid advertisers don't like it. Those videos still get paid from Premium views.

Youtube premium users on average give creators more revenue per view than non-premium users. 55% of the premium revenue is split between the creators you watch.

Monetizing your marketplace monopoly with 45% rents is even more egregious than the App Store which people complain about at 30%.

In fact, it might be the highest monopoly tax in all of tech. Even Spotify only takes 30% from the same musicians who post the same music videos on each platform.


Streaming video requires excruciatingly expensive infrastructure. It's one of reasons why there are no competitors to be seen.

It requires less expensive infrastructure than AI, and AI has tons of competitors.

YouTube simply enjoys a classic network effects monopoly, and that’s why their margins are high compared to any other business in the S&P500.


There will be no competitors if one of the player in the field does it for free for enough time. We'd call that "dumping" if it was a manufacturer.

It would be dumping if they took much less than 55%, but they actually do make profits so its not dumping.

YouTube has been in the red at least until 2010 under most estimations.

For reference that's around the point Vimeo started pivoting to different strategies and blocking long content as they couldn't pay for the infra.

That's also around that time that Dailymotion went down the pipes with the French gov stepping in to save the remains.

YouTube thrived from there as creators and advertisers had nowhere else to go at that point. That's the dumping part.


Video especially with high-bit rates is most expensive medium to deliver and store. Well, I suppose Youtube could move to model where they charge for creators for both of those and drop their cut to 30%...

YouTube pays creators more for each Premium view.

I don't see why it's the issue. Youtube infrastructure is not cheap.

Well, it is not cheap if we look at the massive server racks they have, but in the scale of the world? Watching 1h of a video probably costs them like $0.00001 or something equally minuscule.

How would you see it if your phone company spammed your calls and SMS and offered to remove the annoyances for some random fixed fee that is not tied to your usage of the service ?

If we care about Youtube's infra, the expected business structure should follow that assumption.


> that is not tied to your usage of the service

Could you explain this more ?, i'm sure i only get Youtube Ads when watching videos, which is "usage of the service".


You pay the same fixed Premium fee per month whatever you do with YouTube.

You can quit YouTube for weeks or watch it 22h every day, you still pay the same. Same way you can exclusively watch non monetized streams or only watch top monetized creators, you'll be paying exactly the same.

The only difference will be how much YouTube gets to keep.


> You can quit YouTube for weeks or watch it 22h every day, you still pay the same

This has always been in subscription model, like mobile data plan, or exclusive club membership. I won't argue if it's good or not, just saying it has been a thing for a long time.

> you can exclusively watch non monetized streams or only watch top monetized creators, you'll be paying exactly the same.

Well, the server do not care if the video's creator is paid or not, it still has to store the same data, and you have to pay for it.


Maybe some never experienced an ad because they have been using an ad blocker since before ads on Youtube became a thing?

Or used YouTube a lot in the years before it all went downhill.

Well, we grew up in the Great Pirate Era

Why don't you use ad blockers?

Don’t exist on Apple TV box.

I likewise don't use one on my apple TV, but my friend recently told me there are proxy apps for Apple TV which use DNS-based ad blocking and which can get you the US Netflix library while abroad.

Absolutely. I've been too lazy to do anything like this.

I will happily pay YouTube for just one feature: dub all content I see into my native language.

Oh, I'm not against paying for a service. I'm willing to pay more, but that's the issue: companies will happily sell you their basic enshittified product and never provide you with an option of actual good usable one.

I do feel entitled to unlimited video streaming for free. Since the invention of bittorrent there is no need to have a client-server middleman for distributing large files like videos.

If the bandwidth bankrupts them, then boo hoo. They take advantage of network effects so no one can go anywhere else.

Don't feed the bears. That's what I say


The new thing that YouTube Premium includes is the one button press to skip over "commonly skipped parts of the video"- typically the in-video promotions. This just showed up last week on my nVidia shield connected to my TV. So finally there is a way to remove ads for real. It would be nice if it did it automatically.

The creator is getting paid more from my Premium subscription, so I definitely do not want to see their own ads.


YouTube music being included effectively replaces an additional music streaming service. From that perspective the family oriented plans in particular carry a lot of value.

My gripe with Youtube Music is that the bitrate quality of their music is lower than Tidal or even Spotify. YTM audio files that are actually on Youtube will only stream in 128kbps.

I’m surprised to hear that. I just switched from Spotify to Youtube Music and found the audio quality to be way better, even though I had Spotify set to high.

I imagine most people have the same value prop I do

1) I watch youtube more than any streaming service

2) I really really value not having ads in my life

So the price for ad-free youtube really seems phenomenal. None of the other features really matter to me - ad free dominates all value discussions.


> 1) I watch youtube more than any streaming service

I would amend that to say "any *other streaming service". To me Youtube provides more and better content than the other streaming services, and I don't think people should balk at $14 for youtube when they happily pay that for netflix, disney+, hulu, or spotify.


On top of removing ads and giving you a couple extra minor features, it also has a way better rev split with creators (last I heard). Half of the sub gets divvied up to the people you watched that month, portioned out via watch time.

You also get youtube music, instant skipping over sponsor sections, and the ability to play videos in the background

Aside from Music, these are all negative features that are valuable only because YouTube is so obnoxious.

I'm in vehement agreement with parent to be honest. "We'll stop spitting in your soup if you pay us extra" isn't a nice value proposition.


"We'll stop spitting in your soup if you pay us extra" isn't a nice value proposition

so you want people to freely watch videos without paying anything or watching ads ???

how this works then, creator need to be paid, bandwidth need to be paid, infrastructure is not cheap

it is a nice value proposition, if its not somebody would already make a better alternative that not require those 2 (without paying and without ads)

the fact there is not then its not possible


To stay in the metaphor, wouldn't see some other business model that would allow them to provide the soup to people who order without having to threaten to spit into it ?

lol, there is no spit on it

it is the soup, people free to eat the soup or not

the fact that people always focusing on youtube flaw but never recommend alternative is simply saying that they are the best


That's the hallmark of a monopoly: people can complain about it as much as they want, it won't have any material difference.

"That's the hallmark of a monopoly"

but there is no monopoly ???? are you saying that you simply cant use another website/platform ????

this is ridiculous

if its android/ios then I can understand why its monopoly. but we have bazillion other video/streaming website

Youtube are simply the best, deal with it


It's not a particularly crazy idea that free users get a lesser experience. I'm perfectly happy to pay for youtube since it provides by far the best content and the price is reasonable.

The fact that people can get all of that for free with some minor limitations is fairly generous.


> fairly generous.

Is Google "generous" ?


It feels bad as a consumer, but the alternative is usually worse.

The "stop spitting in your soup if you pay us extra" is really efficient market segmentation. If you don't do that you need to find actual value props that separate the market in just the right way to generate the financials that allow the product to keep going as is. 9 times out of 10 the result is that failing PMs totally fuck up the product and everyone loses.

It's the SSO kerfuffle in a different package - terrible, but the right choice surprisingly often.


There’s an excellent free ios app called “Unwatched” which lets you make playlists, set defaults per channel such playback speed, and lets you play videos in the background. I use it for “podcasts” which are video only.

And you don’t have to log in.


You also get >2x playback speed and higher bitrates on some videos.

Full disclousure: I work for Google, but nowhere near YT/YT music. Opinions are my own and I am actually a customer as I pay for YT premium with a family plan for me, my wife, and our son.

While I agree YT without Ads is great, you also get YT music which is really good and for us it replaced Spotify completely.

Personally, though, I don't have a problem with search (maybe because I set a lot of channels as "do not recommend/show"). Shorts, however, they are really annoying.


> Personally, though, I don't have a problem with search (maybe because I set a lot of channels as "do not recommend/show").

Previously search was just search. It wasn't great, but it wasn't too bad.

Now it shows 5-7 results from actual search (often really bad results).

The next section is "People also watch" which quite often has very passing relevance to what you look for.

Then there are shorts.

Then there's "explore more" which may or may not be relevant to your search, and it has "+N more" underneath.

And then there's the rest of the search which, again, may or may not be relevant to your search at all.

---

I think it was slightly fixed recently, so the results are a bit more relevant, but it still is just ... weird


I use Brave and it's the premium experience already.

You should be paying (or taking some other action) to extricate ads from your life as much as possible.

Paying to remove ads is negotiating with terrorists.

YouTube stays in the dominant position either way, it's not like tomorrow you'll go watch Nebula exclusively (you'd already have done it at this point). They're not providing anything materially, so the amount you pay is bound to nothing except how much you're willing to pay. And how much you're willing to pay depends on how much you're annoyed.

So YouTube's main incentive for this program is to annoy you as much as you can tolerate to optimize the most money you get extracted.


It sounds like you're arguing that YouTube should be free and also ad-free, which makes no sense.

YouTube is expensive to operate. They give me an option of paying by watching ads or paying money. That's much better than my options most other places, which is just to be forced to see ads.


I'm arguing that youtube should be paid for actual features. For instance membership and super chats are clearly labeled as extra content. Member only content is the same.

You pay for a specific thing that is produced by a creator and provided by Youtube. "Pay to remove the ads we're pushing" is none of that.

On Youtube being free, this is their business choice, and also the way they crush the competition and cement a near monopoly on the market. If it was a public service NGO I'd see it from a different angle, but it's not.


Would you then argue for Youtube to take the Netflix path and not provide non-paying users anything?

I'd argue that regulators should have a serious look at the effect of Youtube on that specific market, and if the only solution is the Youtube free tier disappearing I'll be fine with it.

We're in a skewed situation with a near monopoly that only companies at the size of Bytedance can challenge, and I'm not sure why we should see the status quo as something to be protected or encouraged.


> if the only solution is the Youtube free tier disappearing I'll be fine with it.

That'd be something most people wouldn't agree with. People always ask for free link anytime a paywalled article posted.


Absolutely, If premium sorted out all those problems and generally treated creators better i'd have a subscription.

I come to youtube for the *creators*, the actual platform where I have watch history off and use extensions to block the aggressively pushed slop as it currently stands is not something I want to put money towards.

I'm already a patreon to a few creators and have a Nebula subscription; adding it up it's probably slightly more than a premium subscription.


The other useful Youtube Premium feature is the ability to offline download videos to your device. Useful for long plane rides and elsewhere where internet is limited or nonexistent.

Yeah it's hard to compete with Ublock Origin and youtube-shorts-block.

That's why I struggled with Ori and the Will of the Wisps. They subtly broke the movement and some hitboxes, and I could never get the proper hang of it.

Blind Forest though? Chef's kiss


> OOXML is complex because it has to be.

What it didn't have to be is sections upon sections of "this behaviour is as seen in Word 95", "this behaviour is as seen in Word 97" without any further specification or context.

The main struggle for independent implementors was reverse engineering all the implicit and explicit assumptions and inner workings of MS Office software.

> But admitting that would have been hard. Easier to come up with conspiracy theories.

I actually read through a lot of that spec at the time. A lot of it was just lip service to open standards at a time when MS was under a lot of regulatory pressure.


> "this behaviour is as seen in Word 95", "this behaviour is as seen in Word 97"

The office relies on behaviour in windows itself "a lot". Even office mac or office web they made themselves isn't a 1:1 replica of the office on windows.

Let alone describe it as a standard.

"this behaviour is as seen in Word 95" sounds sloppy, but it is indeed the closest they can get.

Or what else can you do? You can't just also ship a installation media of word 95 and windows into the ISO standard, right?


> You can't just also ship a installation media of word 95 and windows into the ISO standard, right?

That's what they almost literally did. The spec is littered with "behavior of this program that has no specification and to see it you need to install it and run it"

And that's on top of re-inventing a bunch of specs in MS-only and MS-specific manner (like dates, for example)


That stuff happens because Microsoft don't know what the behavior is. It's just a bit which forks Word down some ancient code path that nobody understands and isn't properly documented. Given the huge effort that would have gone into producing this thousand plus page specification, is understandable why the spec writers would have given up at times.

I expect most people posting on Hacker News would not be able to write a satisfactory specification for their own software if they are working a large legacy code base.


This is correct and there's no point fixing the bugs because it means the layout of the document will change.

> That stuff happens because Microsoft don't know what the behavior is.

They do. Or they did at the time. They literally had things like "save as Word 95" in their office suite.

> Given the huge effort that would have gone into producing this thousand plus page specification, is understandable why the spec writers would have given up at times.

Given the huge effort to produce it in unreasonable timeline they forced themselves into due to regulatory pressure, sure.

The whole OOXML came about only because some large governments said "well, we don't want to be beholden to black box document formats, and we might want a selection of vendors in the future, so ODF looks like a nice proposition compared to Word, actually".

So it was literally rushed through Ecma. MS submitted 2000 pages in December 2005, the spec grew to 6000 pages over the course of the yer, and got standardised in December 2006. So, only a year to significantly increase the spec and standardize it.

And then it was rushed through the ISO standards track which included things like "Swedish vote declared invalid, accusing MS of manipulating votes" https://www.linux-magazine.com/Online/News/Swedish-OpenXML-V... or "Netherlands automatically abstains from voting due to Microsoft" https://archive.ph/20120711220944/http://isoc.nl/michiel/nod... or "near unanimous 'No with comments' turned into 'Abstain' from Malaysia" https://web.archive.org/web/20090726171905/http://www.openma... or...

Google said it best: https://www.csun.edu/~hcmth008/odf/google_ooxml.pdf

--- start quote ---

In developing standards, as in other engineering processes, it is a bad idea to reinvent the wheel. The OOXML standard document is 6546 pages long. The ODF standard, which achieves the same goal, is only 867 pages. The reason for this is that ODF references other existing ISO standards for such things as date specifications, math formula markup and many other needs of an office document format standard. OOXML invents its own versions of these existing standards, which is unnecessary and complicates the final standard.

If ISO were to give OOXML with its 6546 pages the same level of review that other standards have seen, it would take 18 years (6576 days for 6546 pages) to achieve comparable levels of review to the existing ODF standard (871 days for 867 pages) which achieves the same purpose and is thus a good comparison.

Considering that OOXML has only received about 5.5% of the review that comparable standards have undergone, reports about inconsistencies, contradictions and missing information are hardly surprising.

--- end quote ---

Do not for a second assume that anything about OOXML was done in good faith. Well, apart from the thankless work that people assembling the standard did.


> They literally had things like "save as Word 95" in their office suite.

And what do you think that setting did? Forked execution down an alternative no longer maintained codepath instead of the rewritten version that wasn't quite compatible.


Which shouldn't be in an open spec...

Ah yes. All the changes made to word just continued to work magically when "forked down an alternative codepath no one knew about".

And if that's the case, why was that specified in OOXML?


> save as Word 95

This is for format only and a lossy conversion.

Line break or whatever finer behaviour isn't preserved perfectly.

Or probably any conversion between different office version is a lossy conversion.

They (microsoft) truly don't know how their format behaves exactly at any time I believe.

Even office for mac or office for web behaves slightly different.

I don't believe they have a true spec (A pixel perfect one) about word document. Or they should use it on their office mac or office web first.


One of the many reasons why any advice du jour on "just use this methodology to make agentic coding produce amazing results" is utter crap.

> The people who wrote PB clearly knew ASN.1.

And your assumption is based on what exactly?

> It was the most famous IDL at the time.

Strange that at the same time (2001) people were busy implementing everyting in Java and XML, not ASN.1

> Do you assume they just came one morning and decided to write PB without taking a look at what existed?

Yes, that is a great assumption. Looking at what most companies do, this is an assumption bordering on prescience.


> Strange that at the same time (2001) people were busy implementing everyting in Java and XML, not ASN.1

Yes. Meanwhile Google was designing an IDL with a default binary serialisation format. And this is not 2025 typical big corp, over staffed, fake HR levels heavy Google we are talking about. That’s Google in its heyday. I think you have answered your own comment.


> a set of rules, roles, and workflows that make its output predictable and valuable.

Let me stop you right there. Are you seriously talking about predictable when talking about a non-deterministic black box over which you have no control?


> Are you seriously talking about predictable when talking about a non-deterministic

Predictability and determinism are related but different concepts.

A system can be predictable in a probabilistic sense, rather than an exact, deterministic one. This means that while you may not be able to predict the precise outcome of a single event, you can accurately forecast the overall behavior of the system and the likelihood of different outcomes.

https://philosophy.stackexchange.com/questions/96145/determi...

Similarly, a system can be deterministic yet unpredictable due to practical limitations like sensitivity to initial conditions (chaos theory), lack of information, or the inability to compute predictions in time.


The topic of chaos is underrated when people talk about deterministic systems, but I think it's at least (usually?/always?) a tractable problem to draw up a fractal or something and find the non-chaotic regions of a solution space. You have nice variables to work with when you draw up a model of the problem.

Maybe someone can elaborate better, but it seems there is no such luck trying to map probability onto problems the way "AI" is being used today. It's not just a matter of feeding it more data, but finding what data you haven't fed it or in some cases knowing you can't feed it some data because we have no known way to represent what is obvious to humans.


From the discussion in the link: "Predictability means that you can figure out what will happen next based on what happened previously."

Having used nearly all of the methods in the original article, I can predict that the output of the model is nearly indistinguishable from a coin toss for many, many, many rather obvious reasons.


Yes, and there is absolutely nothing wrong with that. Living creatures are mostly black boxes. It doesn't mean we don't aim for making medicine with predictable effects (and side effects).

I don't think that's accurate.

Don't most people working in medicine usually have biology and chemistry degrees? Are you saying those sciences are dark arts?


Having biology degrees doesn't make you understand every detail of human body. There are many, many drugs that are known to work (by double blind testing) but we don't know exactly how.

The details of how penicillin kills bacteria were discovered in 2000s. Only about half a century of after its commercial production. And I'm quite sure we'll still see some more missing puzzle pieces in the future.


Yes, but I think we want to know how they work? Not knowing "exactly how" but having a good ballpark idea is not equivalent to letting AI throw stuff at the wall to see what sticks.

Medicine that can either kill you, cure you, or have no effect at any given time for the same disease is quite unlikely to even pass certification.

Do you know why?


That is exactly my point.

Then no one understands your point

Maybe you can try to read other comments below your original comment, as they mostly share the same point and I don't bother to repeat what everyone else has said.

I'll put it concisely:

Trying to build predictable result upon unpredictable, not fully understood mechanisms is an extremely common practice in every single field.

But anyway you think LLM is just coin toss so I won't engage with this sub-thread anymore.


And you should read replies to those replies, including yours.

Nothing in the current AI world is as predictable as, say, the medicine you can buy or you get prescribed. None of the shamanic "just one more prompt bro" rituals have the predicting power of physics laws. Etc.

You could reflect on that.

> But anyway you think LLM is just coin toss

A person telling me to "try to read comments" couldn't read and understand my comment.


> Nothing in the current AI world is as predictable as, say, the medicine you can buy or you get prescribed.

Do you know there are approve drugs that have been put in the market for treating one ailment and that have proven to have effect on another or have been shown to have unwanted side effect, and therefore have been shifted? The whole drugs _market_ is full of them and all that is needed is to have enough trial to prove desired effect...

The LLM output is yours to decide if it is relevant to your work or not, but it seems that your experience is consistently subpar with what others have reported.


> Do you know there are

Yes, I know. Doesn't really disprove my point

> all that is needed is to have enough trial to prove desired effect

all that is needed lol. You mean multi-stage trials with baselines, control groups, testing against placebos etc.?

Compared to "yolo just believe me" of LLMs.

> The LLM output is yours to decide if it is relevant to your work or not, but it seems that your experience is consistently subpar with what others have reported.

Indeed, because all we have to do with those reports is have blind unquestionable faith. "Just one more prompt, and I swear it will be 100% more efficient with literally othing to judge efficiency by, no baselines, nothing".


Non-deterministic does not mean not-predictable.

Quantum mechanics is non-deterministic, yet you can predict the motion of objects with exquisite precision.

All these "non-deterministic boxes" will give the same answer to the question "What is the capital of France"


Yes, but the "exquisite precision" comes from the deterministic parts of physics.

Nah. The only thing we can establish precisely at the lowest levels is probability. We can and do engineer systems to maximize probabilities of desired outcomes and minimize probabilities of undesirable ones.

Frankly I don’t understand how software engineers (not coders mind you) can have issues with non deterministic tools while browsing the web on a network which can stop working anytime for any reason.


Because the failure modes are deterministic. An API can be down, which you can easily plan for, but if it's up and returning a 200, you can reasonably expect it to return what it's supposed to.

Would you say a top tier human developer produces predictable output? I would, in the sense that it will be well designed and implemented code that meets the requirements. Can we guess every variable name and logic choice? Probably not.

> Would you say a top tier human developer produces predictable output?

First you'd have to prove that LLMs can be equated to a "top tier human developer"

> I would, in the sense that it will be well designed and implemented code that meets the requirements.

Indeed. Something LLMs can or cannot do with all the predictability of a toss coin.


> First you'd have to prove that LLMs can be equated to a "top tier human developer"

Huh? Can you elaborate? I thought the claim was that predictable output is the gold standard and variance in LLM output means they can never rival humans.

Please restate if I missed why deterministic output is so important.


Huh? I guarantee you, if you give two different developers tbe exact sane set of requirements, that you’d get two very different programs. Try it. They likely perform differently also, performance- or resource-wise.

Would you still call that predictable? Of course you would, as long as they meet your requirements. Put it another way, anything is unpredictable depending on your level of scrutiny. AI is likely less predictable than human, doesn’t mean it isn’t helpful. You are free to dismiss it of course.


> Of course you would, as long as they meet your requirements.

Key word: "as long as they meet your requirements".

I've yet to meet an LLM that can predictably do that. Even on the same code with the same tools/prompt/rituals a few hours apart.

> AI is likely less predictable than human, doesn’t mean it isn’t helpful.

I'm struggling to see where I said they weren't helpful or that I dismissed them


At one point I had two pages in the top spot on HN: https://mastodon.nu/@dmitriid/114852056319245427

- 20k peak unique visitors

- 162k peak requests

- 56 GB peak data but most of that data was cached by Cloudflare


That peak is a 24-hour cumulative value if I'm seeing it right?

Do you have stats for the peak requests / second?

Sadly, no :(

I'm currently on free cloudflare plan, and I don't think it shows rps. And it doesn't show stats more than 30 days back


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

Search: