Hacker News Reader

Comment by ALittleLight | Hacker News Reader

This seems awfully brief and doesn't have much detail. The tittle says it's a "preliminary" report. Is this the expected UAP report or just a preliminary document about it?

I would want to see summary statistics for each reported incident and descriptions of the most compelling phenomenon. This document raises a lot of questions and answers few.

The document says they have 21 reports of UAP that "appear to demonstrate advanced technology" including moving at "considerable speed" "without discernable means of propulsion." I need more detail! Are these reports among those confirmed by multiple trustworthy sensors and observers? What do we know about these UAP? What speed? Do we have video?

I'm also not really impressed by their idea to use "advanced algorithms" and "machine learning". Are UFO encounters so frequent we need to resort to big data? I do think it's an excellent idea to process recorded radar data looking for UAP signs though.

View on HN

> I'm also not really impressed by their idea to use "advanced algorithms" and "machine learning". Are UFO encounters so frequent we need to resort to big data?

I had that same thought - on first read this report seems to contain summary analysis of 144 UAP incidents. Not the sort of numbers that immediately make me think "I need ML to process those!"

ML/AI/"algorithms" are the buzzwords of the day in the public and private sectors right now.

Agreed; ML doesn't make sense to me in the context in which they present it. Seems like they just crammed in a buzzword there.

Later in the report, it says this:

    One proposal is to use advanced algorithms to 
    search historical data captured and stored by radars

They do not mention ML in that context, but it seems like a more promising way in which they might usefully employ it.

144 _reported_ incidents. They also talk about how some of the cases going unreported. It would make sense to sieve through the radar data (I'd guess all the ATC radars are being recorded, because why not?) to look for those.

Yep that's fair enough. And satellite imagery too I guess.

Most ATC now uses secondary radars. Those generally only pick up aircraft carrying transponders.

Good point. What about the radars used by Air Force to guard the airspace? Do they cover the interior, or only the borders?

Secondary radar == just an RX only antenna?

Transmitter also, but not powerful enough to get a useful return from an aircraft without a transponder at any significant range.

I think one proposal is to have advanced wide-field cameras placed in "locations of interest". This would be much like an optical observatory and would generate petabytes of data.

Do you have more details on this or a source? I’m curious!

This program (FRIPON) https://www.aanda.org/articles/aa/full_html/2020/12/aa38649-... was cited here https://www.3af.fr/global/gene/link.php?doc_id=3695&fg=1 as a candidate for repurposing for UAP detection. I'm not familiar with the legitimacy of the 3af french org, so perhaps should be taken w/ a grain of salt?

Interesting!

You might have misunderstood the point of the ML strategy. The idea is to train a model on explainable event data such as weather balloons or wildlife as collected via various existing sensors. Then take this model and use it on historical sensor data to identify events that do not match the traits of the known events. I feel like this is a pretty reasonable use for ML.

If you're curious, this previously released footage is quite detailed.

https://www.theguardian.com/world/2020/apr/27/pentagon-relea...

Reply on news.ycombinator.com