Hacker News Reader

Comment by graypegg | Hacker News Reader

Unrelated to the root post, but I’m so glad that time is over. There was a ~1year period where it was nearly impossible to have a conversation about solving something that didn’t involve someone scheming about how to reframe the problem as an economic model, and then drown out everyone with some half-baked tokenomics concept as a way to “solve it all!”. It really felt like a step backwards in critical thinking for people that I did (and still do, now that they’ve chilled out) respect for their problem solving.

View on HN

>Unrelated to the root post, but I’m so glad that time is over. There was a ~1year period where it was nearly impossible to have a conversation about solving something that didn’t involve someone scheming about how to reframe the problem as an

...

Like NoSQL some time before that, and GPT / "AI" nowadays.

I wonder if these LLMs have been trained on the DejaNews/Google usenet archive.

Those who have never read Edgar Codd's seminal paper on RDBMS are doomed to repeat the sounds-good-if-you-dont-think-about-it-too-much mistakes that NoSQL makes people sleepwalk into - and this will never end because database-theory is always an unpopular elective for CS undergrads (and a very dry, unappealing, subject-matter at that) - everyone wants to do Graphics or NLP or even Haskell instead - and if that's what CS degree-holders are like think about how even less well-prepared (for formal-modelling, etc) that a self-taught or bootcamp attendee would be: my point being that people with the right knowledge are always in the minority in this industry).

I say this isn't new because on the desktop in the 1990s, both Apple and Microsoft's dev platforms included some form of "object storage" based around their respective OLE/COM systems[1] and before then in the 1970s and 1980s it was people building business-systems on AS/400 (going back to the System 360...) where too many people never had a clue how to persist data on-disk (which was the whole impetus for Codd's paper, after-all) - so I think the 2000-2010 years where PHP+MySQL was dominant - and where most of us cut-our-teeth - was the exception: SQL is hard, relational theory is harder, relational calculus is harder-still (I think I got a B grade in that course...).

Non-experts have a very reasonable expectation that platforms should make it straightforward to save and load structured data without needing to spend a year of CS spending hours on exercises about functional-composition, surrogate keys, and decomposing tables to their 6th Normal Form - the embarrassment here isn't the non-experts you might think I'm dunking on, but it's the opposite: I'm very disappointed in the experts: the database-vendors and the platform-vendors, that they haven't solved the object-graph persistence library design problem.

------------

[1]: OLE/COM was for more than just composition, embeddeding Excel worksheets in Word documents and GUI design-tools: COM/OLE also featured a binary object structured storage API ( https://en.wikipedia.org/wiki/COM_Structured_Storage )

Because I was curious and you didn’t link to it, here is A Relational Model of Data for Large Shared Data Banks by EF Codd: https://www.seas.upenn.edu/~zives/03f/cis550/codd.pdf

Thanks, must read.

Cool, thanks for mentioning it.

Even so, I wouldn't dismiss using economic incentives as a means of solving many kinds of problems - for example, proof-of-work started-off as an anti-spam initiative (which I feel still has potential), and outside of CS/SE I feel that cap-and-trade is viable approach to help reduce our aggregate carbon output, and so on.

That the blockchain "brand" has been (irrevocably?) damanaged by fools and charlatans doesn't invalidate the potential of using economic-incentives to solve problems: both technical and societal.

You forgot about obligatory "actually, using normal proper database would make that solution be far more simpler and faster", and impending silence from whoever proposed another blockchain garbage... or alternatively a slew of comments where the blockchain fanboy slowly learns how DBs work and how the fancy things in blockchain are possible without it for 30 years now.

“Wait… I’m seeing column constraints in this white paper for PostgresSQL you sent me… how do we make this trustless??”

Exaggerating… a little bit

I still think you can use economics to model pretty much any human interaction. The problem is, you can't just slap a token over everything and call it a day – the incentives that drive people are way more intricate than that. On that note, I think many systems will function better without blockchain. Trust was driving the internet (and any other distributed human networks) for the long time – long before Ethereum came in.

Hugs are worth more than handshakes.

You can also model any problem as a hydraulic circuit if you want.

There are other incentives like fun, passion, and anger that will break any economic model.

If HN had some internal-economy model where you were cycling around karma as if it was currency, paying upvotes into new posts to get a return back for your “early upvoting” if they do well, giving you more capital to spread to other new posters… I know exactly how the dynamics of this site would change for the worse. People just post here (free work!) because it’s interesting. How irrational!

There’s some merit to the idea that you need aligned incentives. I’ll give you that. Just doesn’t always need to be directly boiled down into units of exchange.

>There are other incentives like fun, passion, and anger that will break any economic model.

Yes, because those incentives can't be reduced to numbers, which is the only thing the economists, bean counters and ambulance chasers, among other professions, can even understand, let alone employ, in their models.

Reductionism at work.

Post count was a big driver for a while.

There's a kind of musical chairs or hot potato game being played with online trust. Nobody wants to pay to post; nobody wants to pay for moderation; nobody wants to moderate uncompensated for a long period; and nobody wants to stay somewhere that's unmoderated(even if the moderation is a bare minimum no-spambots rule.)

The reason why the internet was functioning on trust was because it had economic arrangements powering it in a more invisible sense: the market demand was for more access to content, not higher quality of content. But we're really facing down the final conclusion of that with AI-generated spam. It can be encoded as hours of 8k video with the photorealistic likenesses of multiple celebrities, your spouse, parents and children all speaking to you: it will still be spam. Either the content has some kind of market price or carrying cost on it or we are willing to be elaborately lied to forever.

This isn't even a new issue to economics - the transition from free to regulated markets tends to be precipitated on the demand side by a desire for quality(safe food, air, water etc.) even if the regulation itself is ineffectual.

Blockchains are expensive in terms of data quantity, but they do get at certain parts of the issue; someone who elects to post content through one is signalling way above the noise floor, and it's only biased in that it self-selects to people who own some of that token(which does not necessarily lead to a one-to-rule-them-all billionaire's dystopia - precisely the opposite, it's relatively easy in a technical sense to spin up a small and short-lived chain and hand out some free tokens if you only need a little bit of security for a limited period of time, but that's not a framing we've normalized).

> reframe the problem as an economic model,

Every social problem can be legitimately solved by throwing more money at it[1]. That is, effectively, an economics model.

Corollary: Every economic solution can be improved by removing blockchain from it :-)

[1] Sometimes you have to throw the money at people who will spread your solution, sometimes you have to throw the money at media to spread the solution, sometimes you have to throw the money at campaigning, etc.

> Every social problem can be legitimately solved by throwing more money at it[1]

What?

That famously didn't work with covid.

I don't know what country you live in, but they definitely didn't try that in mine.

> What?

> That famously didn't work with covid.

I was being haha-serious, not completely serious.

The response to "we threw money at $FOO and it didn't work" is "You didn't throw enough money at $FOO".

[PS. which country threw money at the COVID problem and had worse or the same results?]

What about critcoin: a block chain based smart contract that incentivizes critical thinking.

Now you have a new problem: how to accurately identify critical thinking?

Using AI LLM NoSQL low code 4G 6S chain based solution. Use LLM for summary. Then add a critical comment. Then generate summary again. If summary is influenced by comment, and comment is on topic, then comment is critical.

What definitions are you using for "critical" and "critical thinking" here?

(Not a native speaker) It was a joke. I mean that if summary of [original text + message] is different than summary of [original text], then message impacts original text in a major way. This can be used to differentiate impactful messages from nit-picking. For example, every message on HN can be evaluated for it impact on discussion using AI to found hidden infuencers implanted by an alien civilization or by Russians.

Reply on news.ycombinator.com