HNHacker News
TopNewBestAskShowJobs

rgavuliak

233 karma · joined July 15, 2021

submissionscomments
rgavuliak··on Never write your own date parsing library
> Nobody wants people who can do easy things, people want people who can do hard things.

No, people want people that can provide value regardless of the difficulty. What you're describing is how we end up with not invented here syndrome.

rgavuliak··on Windsurf employee #2: I was given a payout of only 1% what my shares where worth
If anyone knew how to spot unicorns, the industry would be very different.
rgavuliak··on Is Jeff Bezos killing The Washington Post on purpose or by accident?
If no-one cared about opinions you would be fine with having just one newspaper that writes down the bare facts. The whole appeal of people paying for media is because they value the opinions on top of the facts that should ideally come from relevant experience/knowledge.
rgavuliak··on Czech president signs law criminalising communist propaganda
Because Czechia has lived through communism so its propaganda is aimed at a concrete period in history, same with nazism.

Also - law needs to be concrete enough to be enforceable.

rgavuliak··on Italy drags Meta, X, LinkedIn into €1B+ VAT showdown: free sign‑ups now taxable?
Oh the ones that do everything to avoid paying taxes in Europe?
rgavuliak··on Adblockers stop publishers serving ads to (or even seeing) 1B web users
> Publishers have just decided to put all their eggs in one basket

Isn't that just the ecosystem provided by the Big Tech ad platforms? Publishers at some point very much wanted to sell ads directly as the CPMs were 10x of what they are now. Then FB and Google took over as the middlemen, pushed CPMs down, added sophisticated targeting and set the precedence for how things are implemented.

Sure the tech giants could team up with the publishers to take back control but they'd need to share more of their revenue in exchange which I am doubtful they'd go for. Not just because it means losing control but also it creates the hassle of having to deal with a fractured landscape of publishers of very varying technical acumen.

rgavuliak··on Diary of a Startup – the rise and fall of ArsDigita
I too worked for a start up that was a combo of services and tech, it too died because the founders got too greedy and tried to spin off a product out of it. Covid had an impact on the business, but I also believe things could have gone better if we could have stuck to a direction.
rgavuliak··on Nobody has a personality anymore: we are products with labels
Wouldn't it then absolve me of my personal responsibility to reprimand the person?It's not me choosing to hold people responsible, it's just something that happens (or doesn't, depending on what was determined)
rgavuliak··on Few Americans pay for news when they encounter paywalls
My point was that your preferences don't make for a good business model when it comes to micropayments.

What I am telling you that based on online readership patterns, people that read 25 and more articles per month don't make up for enough of a base to be worth monetizing at 2.3 cents per article. These are hard facts.

In this case the market evolution is the proof that micropayments are not feasible. Not one company succeeded running them, but you have plenty of news companies using subscriptions that scaled the approach and made significant revenues. These are lessons learned across hundreds of news sites. With all due respect, your opinion is just that - anecdotal.

rgavuliak··on Few Americans pay for news when they encounter paywalls
Let's look at the economics, back when I was working in the paywall industry you had roughly 10% of users reading more than 5 articles on an average medium.

Let's say that you could get 2 % of these people to pay you $10 per month. With a readership of 1M that would mean you'd get (1000000 * 0.1 * 0.02) * 10 * 12 = $240k per year.

Now let's move that over to blendle, let's say that the average reader reads 1.2 articles per month (since most people only look at 1 article). Let's be super generous here - with a much larger conversion rate - 10 % of everyone buying 1 article per month month on average comes up to a comparable amount 100000 * 12 * 0.23 = $276k per month. And we're being generous here, remember - blendle abandoned this model so it almost certainly isn't equal to the metered model I used in my estimation above even if more users are willing to pay.

Now if you bring down the payment to whatever you're proposing you get to $27.6k or even 2.76k per YEAR. The napkin math is clear that this isn't feasible.

Interestingly enough you'd pay less as a "heavy reader" if you read 10 articles per month (which is a super small % of users on an average medium) with blendle (10 * 0.23 = $2.3) than with a subscription costing $10. But making the users count articles likely isn't great psychology as compared to buy and read everything you want. Which likely is perceived as a higher value offer.

rgavuliak··on Few Americans pay for news when they encounter paywalls
> And nearly the entirety of retail sales

Not really? Retail is mostly driven by normal transactions.

> Aside from logistics, the problem with micro-transactions for news that I can see is that it would encourage only the more sensationalistic news and discourage the more mundane news that is likely more important for people to know.

Exactly - it essentially kills journalism that requires a lot of upfront research or work.

rgavuliak··on Few Americans pay for news when they encounter paywalls
It is essential, but the ones I mention are something that is very hard to do without. While there are alternatives they're not widespread and require significant shift in operating style to roll out. Newspaper content is not a necessity so it doesn't work the same way.
rgavuliak··on Few Americans pay for news when they encounter paywalls
You are wrong. There have been multiple attempts at micro-transactions - and they all failed. One of the biggest was Blendle - https://www.pugpig.com/2023/08/18/why-micropayment-champion-...

Why? why do most B2B companies prefer subscription based pricing? Because it brings in predictability you can run a business on. Micro-transactions only work in utility context (electric, water, cloud compute). You can't run a news company on micro-transactions.

rgavuliak··on Go is a good fit for agents
This reminds me of a talk named - Modern Data Science in Go. That line of thinking went nowhere and many arguments in that talk were misleading.
rgavuliak··on OpenAI dropped the price of o3 by 80%
I don't think the no moat approach makes sense. In a world where more an more content and interaction is done with and via LLMs, the data of your users chatting with your LLM is a super valuable dataset.
rgavuliak··on You Weren't Meant to Have a Boss (2008)
Exactly and wait for 2 weeks to get the correct rights.
rgavuliak··on What “working” means in the era of AI apps
Maybe it's the fact that we've seen VCs and a16 specifically push for highly overhyped things that didn't really come true like WeWork, Crypto?

If we go beyond a16, the last decade has shown us VCs pushing for insane valuations for companies that went under after hiring sprees. We've seen Mark Zuckerberg sing similar odes to Metaverse.

At some point you stop believing the next big thing from a certain crowd.

This doesn't mean I am not exploring how to use AI and even have something I am working on in terms of an idea, it's just that caution is advised.

rgavuliak··on UK tech job openings climb 21% to pre-pandemic highs
I live in Central-Eastern Europe and every time a recruiter tried to get me to move to the UK, the salary was at best on par with what I made over here. Granted, up until recently I've worked remotely for US and other EU countries.
rgavuliak··on UK tech job openings climb 21% to pre-pandemic highs
UK has pretty bad tech salaries especially for companies located in London and adjusting for living costs.
rgavuliak··on xAI to pay telegram $300M to integrate Grok into the chat app
Since Grok is too left leaning despite all of Musk's efforts, he needs to get in some more training data that has an alt-right conspiracy bias.
rgavuliak··on Duolingo CEO tries to walk back AI-first comments, fails
For the team that worked on a feature for month it's the whole world at the time of release. Being mindful that is not the end-users whole world, but just a tiny insignificant fraction is something easily lost in denial.
rgavuliak··on Tell HN: The Hetzner Experience - Invisible Outages
In my first job we've used Hetzner for Data Science work. The we lost servers twice in a year back then.
rgavuliak··on I failed a take-home assignment from Kagi Search
I think it's more of a case for legal and HR being conservative and super defensive. Not sure if you've ever handled a contract with an internal lawyer, but in my experience they often go for crazy suggestions that the other side would never accept for the sake of protecting the company as much as possible. Might be the same here - HR/legal being super protective and the hiring manager not caring enough to fight back.
rgavuliak··on EU ruling: tracking-based advertising [...] across Europe has no legal basis
There's also studies on load times impacting conversion rates (which mostly matters at scale i.e. Amazon).

At the same time - quantifying this is not straightforward and companies mostly ignore committing resources to such activities.

rgavuliak··on Bloat is still software's biggest vulnerability (2024)
I agree fully, most users care about making their lives easier, not about development purity. If you can't do both, the puritanistic approach loses.
rgavuliak··on Knowledge-based society, my ass
I had a slightly better experience on my PhD in Central-Eastern Europe. My supervisor - a politician with little academic results mostly cared about the formal part of my results.

I used the PhD to get into a new field (statistics) just in time for me to get into Data Science as it started being cool. My papers weren't interesting, but I got to try things out on real data and report on them.

rgavuliak··on Stop using the elbow criterion for k-means
How does this reply make sense?
rgavuliak··on Stop using the elbow criterion for k-means
Usually I try to tie the clusters to some kind of a bottom line metric. It's not always possible though.

To be honest I am not a big fan of segmentation as is, since it results in a set of subjective decisions when both deciding number of clusters and their interpretation.

An alternative is DBSCAN which comes up with natural clusters, but its also not 100 % foolproof as it has a group of unclusterables which at times ends up too high.

rgavuliak··on Stop using the elbow criterion for k-means
Elbow method was never working well in practice. In my over a decade of experience I am yet to see a significant separation in the chart.
rgavuliak··on Grok3 Launch [video]
As a person who works with data and has done both consulting and product building in Data Science and lack of domain knowledge is what makes or breaks the end result. Too often do technical people think they know better and then build mediocre solutions that don't get used.
← PreviousPage 2 of 7Next →