HNHacker News
TopNewBestAskShowJobs

carlossouza

488 karma · joined March 22, 2022

submissionscomments
carlossouza··on Threads, an Instagram app
Advertisers are persuaded by underpriced attention.

In the beginning, every social network tends to offer that. (Some call it attention arbitrage.)

To drive a critical mass of users, social networks offer incentives to content creators so they flock to them and start producing content there.

The main incentive is the aggressive free distribution of organic content.

Once the social network matures, it reduces the proportion of free content in favor of paid content. And the attention arbitrage goes away.

carlossouza··on How to Do Great Work
Although I enjoy reading his essays, yes, they tend to be longer than usual.

I wonder whether he uses an editor to provide constructive feedback before publishing it or just writes and clicks "publish."

Interestingly, I found that point missing: people who do great work usually have editors/mentors/advisors to help them along the way.

carlossouza··on Building Boba AI: Lessons learnt in building an LLM-powered application
Remember that the most downloaded app in Apple Store’s first year was this infantile thing called iBeer… it actually hit the mark of $20k of sales/day

This was an extreme bizarre example, but most mobile apps from the first wave were also pretty dull

Give it some time. I think something similar will happen: most LLM apps from this first wave will be forgotten soon, and when people shift the focus from the technology back to the users and their real pain points, the good stuff will surface

carlossouza··on CLI tools hidden in the Python standard library
This is one of the best ChatGPT use cases: creating regex complex patterns
carlossouza··on Visa to acquire Brazilian startup Pismo for $1B
Sure: one was a head trader for a specific prop trading desk, and the other was an executive director.

The GTM: the end customers are hedge funds/institutional investors, which are pretty hard to approach directly. So, we developed a B2B2B: we sell to large brokerage firms, which resell to the end customers.

carlossouza··on Visa to acquire Brazilian startup Pismo for $1B
Exactly. I started one. 2 of my cofounders had 15+ years (each) of experience in Credit Suisse and Santander. They knew everything and everyone.

Without them, it would be pretty hard to validate the idea (and basically impossible to sell).

The legal stuff is unbelievable...

carlossouza··on How to make your scientific data accessible, discoverable and useful
Metadata is necessary but not sufficient.

Imho there aren't enough tools to discover scientific data.

carlossouza··on How to make your scientific data accessible, discoverable and useful
The article doesn't fully answer its title, especially the "discoverable."
carlossouza··on The idea maze for AI startups (2015)
Interesting read. I'd argue the most successful AI-based products are the ones that settle for 80-90% accuracy and “Create a fault-tolerant UX.”

Then, the question becomes: how to create a great fault-tolerant UX?

There are some nice recent cases... Github Copilot is one...

carlossouza··on The only acceptable rating to leave an Uber driver is 5 stars
I'm not a psychologist, but I'm sure one can explain this insight:

Whenever I'm rating an Uber (or any service), the number of stars I leave is often more correlated with how I feel at that particular moment (regardless of the service) and less with the service itself

If I'm having a good day or in a particularly good mood, everything is 5 stars, regardless of the service... and vice-versa

carlossouza··on Bootstrapping Atlassian: $10k In Credit Card Debt To $20M In Revenue In 5 years
> ... but that product had terrible support and documentation. So they decided to provide documentation and support services for it, charging customers $360 to solve their issues. However, they faced two problems: First, their documentation was so good that customers rarely needed to call them...

Interesting to see how documentation & support are such pain points to sell enterprise software, and yet how few companies manage to differentiate themselves by making truly great documentation & support experiences. Stripe is another good example

carlossouza··on Happiness is a reward from our ancestors
https://youtu.be/4q1dgn_C0AU

“The surprising science of happiness”, from Dan Gilbert, is one of the best TED talks of all time (over 4M views). In his amazing & funny 20’ presentation, he explains how our brains are wired and how it relates to happiness. The article reminded me of this great content

carlossouza··on Show HN: I built a website to buy and sell used clothes in Indonesia
Nice, well done! Btw, there's a public company in Brazil that does exactly that: https://www.enjoei.com.br/

Their GMV is ~US$ 200 million/year, and their revenues are ~US$ 30 million/year.

carlossouza··on Ask HN: Have any of you had success trading the stock market?
You are right for us, non-professional traders. But there are professional traders who devote their lives to this activity and are successful.
carlossouza··on When Silence is Power (2013)
Easier said than done. To become silent and refrain from commenting "pseudo-events" is easy; to become silent and refrain from clicking in a hyperreality clickbait title is much harder, if not impossible. These days we speak not only when we write/comment online but also (and most importantly) when we click. And with every click, we feed the current broken system. Not clicking is hard.
carlossouza··on Harvard’s new computer science teacher is a chatbot
I think most people do not pay for teachers' quality. People pay for the diploma's brand. (Which is a proxy of the probability of landing a good job after graduation.) For that use case, as long as the chatbot teachers don't degrade the university's brand name, I think most people won't care. They might even actually prefer it. (I graduated from the top engineering university in my country and I'd certainly prefer ChatGPT over the vast majority of the TAs I had).
carlossouza··on Ask HN: How much would you pay per test case?
I think a reasonable approach to sell would be try to convince the buyers that your QA-as-a-Service would replace QA employees for a fraction of their cost. Thus, pricing would be a fraction of the cost of these employees. I think it's a good idea. Btw, you can pretty much validate it entirely before getting started: just approach companies with mid-large QA teams and try to convince them of the value prop (high-quality QA at a fraction of the cost).
carlossouza··on Show HN: Build a personal knowledge graph from the content you consume online
Nice tool. How do you plan to drive awareness/growth?
← PreviousPage 4 of 4