Dropbox to reduce global workforce by about 16%, or 500 staff

This is what got me to get rid of Dropbox. Used them for free for a decade, and they pushed me out by limiting the number of clients. I looked for a paid option for $10-$20 per year since I basically only need a few GBs of space, but they don't have that. So figured I might as well move to my own Nextcloud instance. It's actually still free since I'm using Oracle Cloud's free tier, though of course there's my admin time, but I actually sort of enjoy that.

> You'll also have some time TODAY AND TOMORROW to wrap up and say goodbye to colleagues.

As an European, I don't think I will ever get used to the fact that in the US people can be told "go home, you're fired" like that, effective immediately.

I think you are still trying to treat it as an engineering analysis.

The question isn't "who is no longer necessary".

The question is "do we think we can cut 500 roles (Y/N)"

The 2nd question is a reasonable and useful business assessment. The first question is not relevant and has no real answer.

I imagine things like this are underway: https://github.com/jjuliano/aifiles

Honestly I think it's actually a pretty fantastic development if that's the direction. I don't use Dropbox much, nor have I used aifiles (due to sending all of your data to OpenAI) - but the idea of being able to not have to manually and tediously look over terabytes of files to completely reorganize all of your files into a better directory hierarchy, and tag each files with meaningful labels, etc sounds phenomenal.

Obviously there are some implementation details for this to be not awful - for example: 1) only local models for local data, 2) making the changes on e.g. ZFS (to allow rollback) or as some type of optional 'overlay' view to switch back and forth from the original to ai-organized, etc. and 3) having thresholds and logic for what may be considered 'duplicates' to be removed, and how to better compress data

As for the de-duplication and processing: this could be very good for dropbox in that, e.g. if a person wants to completely re-encode all of their image files or video files with AV1, the resulting data could be cut in half or more - which saves Dropbox storage space. After which, neural perceptual hashing could be done on all of the files and a threshold of similarity could do de-duplication on a perceptual basis (for example, keep the bigger size file that is 99% similar to a 2x downsized version, and re-encode it). User preference to keep things like tiff files completely intact or any other lossless encoding of their choosing could be good options as well

There's definitely a strange disparity between the computation cost for deploying a decent model to do this compared to the storage cost - but if a (perhaps even non-LLM!) small model is created to be able to plow through data at fast rates could be deployed it may make sense.

Or perhaps the type of semantic compression that LLMs do are of interest for making a new type of lossy compression algorithm of which Dropbox is interested. It is certainly interesting that even the largest LLMs we are aware of contain only a few hundred GBs of weights, but have information on various topics from Pokemon to electric engineering, to cellular biology, to cake decorating - yet one instance of the best compressed pubmed database I can think of (only containing the small fraction of cell bio and med part of the LLM - perhaps 2% of it) is like 60GB with Zstd compressed (yes, yes, I know lossy vs lossless - but still) - which would be probably more like 1/2 or 1/4 of a typical SoTA LLM. So clearly lossy compression via including semantic info is hugely beneficial over typical lossy compression techniques (and perhaps one could argue, what often lossy compression algorithms try to model in the first place, in a very rudimentary way).