826 karma · joined May 1, 2009
also i hear your point on swe roles and don't disagree
And unfortunately one not exclusive to OpenAI. Anthropic credits also expire after 1 year.
that's not to say the goal should always be to scale or become as big as possible, but it's interesting to think about.
though, to be fair he said it would take a year to do so: https://x.com/paulg/status/1830300111232188626?s=46&t=8sSeDI...
I also understand people believe that there is a qualitative difference between the venom and synthetic, but it's the same compound at the end of the day.
Also, zoasty explains in the pastebin that he had a lot of help setting hoot up, so I don't think the expectation is for others to be able to easily recreate what he did.
https://www.twitch.tv/sgqfmfunk
More info on the channel:
"And yet, until very recently, Lyft had limited control over that navigation experience, which was built using Google Maps. “For our entire existence, we couldn’t fundamentally change that product,” says director of project management Ben Schrom."
I've worked in the mapping space for more than 20 years and I can tell you one of the biggest annoyances with using third party mapping data is that you are beholden to their cadence and tooling for making updates to the map - you have virtually no control over the data. This is a big reason both Google and Apple built their own map.
The benefit of using OSM is that you are provided decent tooling and a relatively streamlined process for making updates. Map data quality is incredibly important, especially for navigation and it's really hard to manage because the map is always changing - it's not a static thing.
“One possible valid bound for the timestamp of Event B is a value that is greater than the timestamp of Event C but less than or equal to the maximum possible timestamp. This ensures that Event B happened after Event C but not necessarily after all other events.
Alternatively, we could set a valid bound for the timestamp of Event B as a value that is greater than the maximum possible timestamp but less than or equal to the minimum possible timestamp. This would ensure that Event B happened after all other events but not necessarily before all other events.”
The second solution doesn’t make sense to me.
Idea taken from: https://twitter.com/stanislavfort/status/1599417463199830017
If TomTom was planning on conflating OSM data with other sources then in theory it hurts their business model because they will potentially have to make their entire map (or large parts of it) freely available in OSM.
If they can somehow set up clean room conditions and keep all their other data sources physically separated from the OSM data then it might be avoided, but then how do you do the map conflation? And that would certainly impact how quickly they can make updates.
Maybe they came up with some clever way to avoid having to contribute data back to OSM and that’s what OP means by “steal.”