HNHacker News
TopNewBestAskShowJobs

afiodorov

784 karma · joined June 21, 2019

Likes Computer Science, Mathematics, AI, ML
submissionscomments
afiodorov··on Show HN: IMDb SQL Best Movie Finder
done
afiodorov··on Ease of maintenance is a feature
More software is written than kept. It's harder to write useful software than to configure CI/CD. The latter is a problem that has been solved before, whilst chances of any software codebase being useful enough that it is even worth maintaining are very low.
afiodorov··on Show HN: IMDb SQL Best Movie Finder
A 100 mb is not much - about 2 mins on youtube
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Bollywood is the second largest film industry!
afiodorov··on Show HN: IMDb SQL Best Movie Finder
First load takes some time, it is downloading a parquet that's a 100mb and has more than 1.5 million titles of series and movies.
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Ah 7 might have been a typo. May be it was meant to be *7? Then that's exactly the Laplace estimator. We start with a prior of 1_000_000 votes of 7.0 and then adjust as we accumulate real votes.
afiodorov··on Show HN: IMDb SQL Best Movie Finder
The data was produced using this script: https://github.com/afiodorov/imdb-sql/blob/main/notebooks/im...

It is on hosted on http://www.imdb-sql.com/imdb01-11-2024.parquet

In fact the main reason this project exists is

a) I wrote a jupyter notebook ages ago that'd join the raw data into a queryable form.

b) I committed it and forgot about it after the initial viewing of top movies/series.

c) For the Halloween I wanted to find a good-rated horror movie, a genre I don't watch much.

d) I found my notebook but it was a drag to get it running again, first pandas would keep throwing OOM errors so I had to migrate to polars. Secondly I had to find a spare working laptop since iPad is my primary off-work computing device as of late. Finally the schema is not so intuitive - I need to look up things like region/language codes.

I want something I can just leave unattended but still be able to use a few months later without going through a dev-env set-up. A static website with all the data is one potential way of avoiding maintaining it as it should keep running, just like an exe would keep running on Windows in the past.

I avoid maintaining personal projects as my software job requires me to maintain software and I don't find it enjoyable to do the same in my spare time.

afiodorov··on Show HN: IMDb SQL Best Movie Finder
That's awesome. Care to explain? Looks like Laplace estimator of some kind?

P.S. Query sharing is supported:

https://www.imdb-sql.com/?query=SELECT+*+EXCLUDE+%28titleTyp...

afiodorov··on Show HN: IMDb SQL Best Movie Finder
Whole dataset http://www.imdb-sql.com/imdb01-11-2024.parquet loaded (100mb, 1.6 million titles).

It has these columns: https://www.imdb-sql.com/?query=SELECT+*+FROM+%27imdb01-11-2...

It is produced using https://github.com/afiodorov/imdb-sql/blob/main/notebooks/im...

From this data: https://developer.imdb.com/non-commercial-datasets/

The age rating is not part of the imdb non-commercial dataset.

afiodorov··on Show HN: IMDb SQL Best Movie Finder
doesn't appear possible from https://developer.imdb.com/non-commercial-datasets/
afiodorov··on Show HN: IMDb SQL Best Movie Finder
I initially loved looking for obscure stuff, e.g. setting region to soviet union. It surely is the case that 99% of the users want 10% of the data at most. I'll have to work ability to select the file and download & cache it only if the relevant query is asking for it.
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Shouldn't be hard to find since the repo name on github is same as the domain :).

I just run npm run dev locally. You'll need the parquet file of the data in public/ though, can get it from http://www.imdb-sql.com/imdb01-11-2024.parquet

afiodorov··on Show HN: IMDb SQL Best Movie Finder
It'd be trivial to ask LLM to generate the query; but since it's a client-side app there's nowhere to store the api key - so each user would have to supply one, which is a bit of an awkward experience.
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Browsers are becoming OS'es :).
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Will think about improving the ui in this aspect...
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Thanks for the suggestion, will try to improve soon.

If anyone with actual css experience wants to open a PR in the meantime I'll be very happy!

afiodorov··on Show HN: IMDb SQL Best Movie Finder
Good spot, will deduplicate in the next iteration.

However titles are repeated often due to the region/language variations.

afiodorov··on Show HN: IMDb SQL Best Movie Finder
Sorry about that. I prefer for my UI to be discoverable.

P.S. We can share links too https://www.imdb-sql.com/?query=select+1

The SQL flavor is https://duckdb.org/docs/sql/query_syntax/select

afiodorov··on Show HN: IMDb SQL Best Movie Finder
https://github.com/afiodorov/imdb-sql
afiodorov··on Show HN: IMDb SQL Best Movie Finder
It's slow on the first load as you download a 100mb parquet file with all the IMDb data. It'll be cached on the second load though (using browser's indexeddb to cache).

Not sure what you mean by scraping in parallel.

afiodorov··on Accountability sinks
I've observed a phenomenon in corporate accountability resembling quantum behavior:

1. Macro level: Departments claim broad accountability.

2. Micro level: Pinpointing task ownership causes accountability to vanish.

3. Indefinite states: Ambiguous tasks linger without resolution.

4. Entanglement: Dependent tasks inherit this ambiguity.

This creates a system where responsibility exists in superposition, tasks remain unresolved, and accountability becomes increasingly delocalized.

afiodorov··on Language is not essential for the cognitive processes that underlie thought
> What this tells us for AI is that we need something else besides LLMs

I am not convinced it follows. Sure LLMs don’t seem complete however there’s a lot of unspoken inference going on in LLMs that don’t map into a language directly already - the inner layers of the deep neural net that operates on abstract neurons.

afiodorov··on What P vs. NP is about
Idea is to add enough constraints to the function so that inverting it is non trivial (where inverts finds ANY input). Like the video said inverting multiplication with the output of 15 could yield 15 = 15 * 1, so just make the function f(a, b) = a * b where a > 1 and b > 1 for a more interesting RSA-breaking case.
afiodorov··on NotebookLM's automatically generated podcasts are surprisingly effective
Reminds me of GTA3 radio - can we retrofit this somehow? I miss driving around mindlessly and now we can get actual quality podcasts too.

I wonder which successful game will make use of AI generated content next.

afiodorov··on Ask HN: Device for Music and GPS
No, all of them do
afiodorov··on Ask HN: Device for Music and GPS
probably not the answer you’re looking for but iPads match the description (usb-c headphones for the newer models)
afiodorov··on SpaceX Astronauts Begin Spacewalk, Putting New Spacesuits to Test
No, in this case it is much more pragmatic. Elon has commented on this aspect himself. When he started he thought of just buying rockets and related equipment from Russia on the cheap but quickly found the exuberant prices states want to charge. He then made a first order approximation of how much it’d cost to just manufacture things he needs and it was an order of magnitude lower - thus he bit the bullet and started doing just that. Had the states let him buy things at a fair price may be the controlling of the stack would not have happened.
afiodorov··on Claude's API now supports CORS requests, enabling client-side applications
The browser runs the code, which is open-sourced. You can look at the network tab of the debugging console or isolate the browser completely by sand boxing it. I agree that UX can be improved but that’s more on the browser side.
afiodorov··on Claude's API now supports CORS requests, enabling client-side applications
I made two one-page react apps

https://www.livetranslate.net/

https://www.subsgpt.com/

Second one is more refined but both are functional.

I was pleasantly surprised somebody made a YouTube tutorial in Japanese about the latter https://www.youtube.com/watch?v=8gAkvZYayEc - feels like retro internet where people share things on their personal webpages.

Curiously the first one also landed me a contracting opportunity for a company that wanted to add live captions to their product and we went live with my help.

I also made a decentralized twitter dapp ages ago but AI apps definitely have had more interest.

afiodorov··on Claude's API now supports CORS requests, enabling client-side applications
I absolutely think that OpenAI or Anthropic should provide such integration. It’s very similar to how Apple Pay centralises your subscriptions and makes payments secure and simple. Would be nice if AI labs had an equivalent portal where each authorized app gets its own key and I can cancel any time and control my spending. Finally that might enable some kind of monetisation if OpenAI or Anthropic give developers a cut, e.g. 10% mark-up that goes to the authorized app.
← PreviousPage 3 of 5Next →