Show HN: Spotify streaming GDPR dump local analyzer | Hacker News Reader

Show HN: Spotify streaming GDPR dump local analyzer

github.com

104 pointspdubouilh23 comments

And ChatGPT is good at adding comments along the way in the code, so when they do a new dump from GitHub they have way better training data

I already find original ChatGPT (free version based on GPT-3.X) being "smarter" than Bing's one based on GPT-4. The latter much easier gives up just saying that it could not find anything, while ChatGPT replies based on its "knowledge" even on quite advanced science questions. Having a LLM trained on content generated by other LLM may be a way for disaster in future: filtering true from false (but convincing) information on scale would be hard to say the least. In light of recent troubles with Internet Archive, AI companies, like OpenAI, are of the beneficiaries of the Internet history "made on Earth by humans" :)

I think Bing Chat’s safety features makes it very neutered. Lately I’ve been enjoying using Bard more due to its longer answers by default.

You want longer responses? One of my problems with chat gpt is its tendency to use too many words for no good reason. Most of it is just filler imho

You can just ask it to give concise responses and it will play along.

Until the instruction falls out the back of its memory and gets lost in the summarisation! :)

You cant prevent an LLM from having a hallucination.

The current state of the art is retrieval + summarization, all that knowledge it was trained on still exists. When performing a search having no reference for the knowledge is a decent signal that it may not exist at all and you may be talking to a liar.

You can though, by designing your prompt against it and using very low temperature values so it's more deterministic.

Well, this could be close to what I dreamed off when I did the GDPR takeout two years ago! Haven't used Spotify since, so the data should actually still actually be up to date. :)

Never used Spotify on a mobile device though, so the location scripts will likely not be interesting at all.

Oh hell yes! I downloaded my GDPR dump a few months ago with the intent of analyzing it like this but never got around to it. Gonna fire this up now.

On the topic of ChatGPT code projects, way I recently made another evening side project[1] using ChatGPT. I found a really nice pattern to use with it is to use one commit per ChatGPT iteration (including commits where it breaks the program, just don't push to main until it's good again). And in each commit, I store the full prompt or reply I said to ChatGPT as a prompt.txt[2]. I'll probably tack it onto the commit description next time for ease of reading. But other friends have found this really useful to be able to see exactly how I+ChatGPT evolved the software with each commit, and I can look back and reference useful prompting patterns I used.

[1] https://github.com/dag10/timelapse [2] https://github.com/dag10/timelapse/commits/e77d11baaaf4e2a5f...

There's an iPhone/Android app that does something similar: https://stats.fm

Feels abit creepy they have the "mood" and location in a database.

Not sure what you mean, Spotify tracks have long exposed "features" that include various metrics to judge mood. https://developer.spotify.com/documentation/web-api/referenc...

Also it's pretty apparent since Spotify suggests different playlists depending on time of day, weather, etc.

Do they only judge a song's mood, or do they judge the listener's mood and store that alongside the location?

Each song gets analysed and assigned ~10 audio parametres. If you tend to like tracks with, say, high danceability in the past X days, you're gonna get recommended more tracks with high danceability.

If you listen to a specific genre, it's gonna work great. If you tend to listen to unrelated tracks, your recommendations are gonna be shit. If you listen to a couple of albums that are really not what you listen to usually, it's gonna take you a while to get rid of similar tracks from your recommendations.

So to answer your question, it's more of a snapshot of the mood of the tracks you've interacted with recently, not your mood specifically.

I guess I didn't realise that they kept full IPs for such a wide range of time - I was kind of hoping they were anonymizing them after some time...

What's this? What does GDPR have to do with it?

I’m guessing Spotify users in the EU can request an export of their data that Spotify holds. And this tool is for analyzing that data export locally.

That is correct.

For those wondering where to request the data export. The data request form is located here: https://www.spotify.com/account/privacy/

made a lot more sense when i looked up what endsong_*.json was. you can download your data (prob relates to gdpr requirements i guess) here https://www.spotify.com/us/account/privacy/ details as to what is included in json dump is noted here https://support.spotify.com/us/article/understanding-my-data...

Really wish there were more of these kinds of tools to make sense of GDPR dumps.

make some? point out some gdpr dumps you'd like tools for, what they should do?

Reply on news.ycombinator.com