Have Microsoft charge people $ for their repos, and then take their code to train their LLM for more $.
And they can use the surplus $ to fund open source projects to produce more code to train their LLM for even more $, and reduce their taxes thanks to the charitable donations.
That's kind of a narrow take; the mainstream may be directed towards a good thing and just not have the depth to draw a benefit, its attention being superficial and fleeting.
E.g. the Pavlovian dog metaphor is quite a mainstream trope, but doesn't it carry an important message nevertheless?
If anything, I would say that fleeting takes and offhand dismissals are what determines and solidifies the mainstream's superficiality.
>> I don't see it as democratic or democratising. TBH the knowledge is stored in three giga companies
It can appear democratic while access is allowed, but if it can be revoked at any moment for any reason (it is private companies, after all, that own the AI playgrounds), then the illusion will shatter.
What is more, excessive reliance on AI creates skill deficit rather than skill surplus, and promotes dependence on AI. Wizards that are nothing without their magic wands, in a way.
This may not stand out today, but give it half or one decade, when the next generation won't have a pre-AI skillet to fall back to, and the seams will become all too apparent.
I already notice that my brains tends to resist thinking on hard things, defaulting to gpt-ing or perplex-ing things. Similar feeling I had when I bought a car with parking sensors - I almost immediately lost skills of parking my older car which doesn't have those. I had to re-learn it.
Nice coincidence, I was thinking about the sale yesterday.
IIRC, a while before the sale Whatsapp tried to introduce some meagre subscription, on the order of dozens of cents, which got a lot of backlash. Then, a bit after that, it got sold.
The servers don't pay for themselves, and if the user base wasn't going to pay for use, money had to be manifested in another way.
WhatsApp on iPhone was initially $1 to download, but had frequent sales. WhatsApp on other platforms was free to download woth a $1/year subscription... But subscription enforcement was uneven.
I started in 2011, and the subscription language was present, but there was no mechanism for payment. Then we put payment into Android, but frequently would extend all subscriptions. At some point the iPhone model flipped to match the rest, but if you had registered with iPhone before the switch, your account was set to lifetime.
I don't know the timeline, but towards the end there was a small list of countries where we would actually enforce loss of service for about a week when the subscription ended. After a week, we'd extend the subscription for a while anyway, because it was probably hard to pay (we tried to pick subscription enforcement countries where payment was readily accessible, but lots of people don't have a compatible mode of payment even if they have the means to pay)
We were told the company was cash flow positive, the public GAAP numbers look bad, but a large part of that is stock based compensation; a small part is accounting treatment for the lifetime accounts.
Also, it's important to note that the acquisition happened before real time voice and video calling launched and running servers for that was expected to be expensive.
> Speaking at the DLD conference in Bavaria, Jan Koum confirmed that the $0.99 annual fee will be scrapped, effective immediately. Previously, WhatsApp had been free for the first year, with the fee charged for every subsequent year. Long-term users of the iOS version were given free use for life, as a thanks for paying a fee to download the app when it had a one-off charge.
Have Microsoft charge people $ for their repos, and then take their code to train their LLM for more $.
And they can use the surplus $ to fund open source projects to produce more code to train their LLM for even more $, and reduce their taxes thanks to the charitable donations.
Everyone wins, right?
Thankfully we still have Codeberg.
reply