admin said in #3222 4d ago:
Hey anons, this week we have a big delivery: the sofiechan newsletter. It seems to work. We're shipping it. Head to the "you" page and click the endorse icon on the newsletter question to subscribe, and you should receive something in the hour after Monday 16:00 UTC/9:00 PT. Even if you subscribe late, you'll still get Monday's newsletter if you subscribe by about Thursday I think.
It's a pretty simple deal: the top five threads from the last two weeks minus anything that's already been featured in a newsletter. So if you post on Saturday and it doesn't blow up until Wednesday, you'll still be featured in next Monday's newsletter as a late bloomer. Basically, it should just work.
Hopefully it all works smoothly. We've tested it a bit, but as you might imagine it's hard to realistically test a periodic newsletter. There may be delivery issues, rendering issues, etc. Please dig it out of your spam box and tell your mail provider it's not spam. Report any bugs in this thread or to me by whatever means. We'll work out the kinks and get it ready for primetime over the next few weeks.
We are always happy to hear your feedback about any aspect of the system, especially at these early stages. Let us know how we can make this newsletter great for you.
Meanwhile in other progress, I completely rewrote the tag system backend. I hope you don't notice. It is now in a form where where it will be much easier to integrate more advanced features like autotagging based on linguistic analysis. I've got a plan and I think it's going to work. The autotagger will not only extrapolate from our tag use patterns to suggest tags based on post content, it will be capable of refining the boundaries of these tags and even autonomously discovering and managing new ones (which we can then rename and use as we see fit). It will generally defer to us, but will attempt to steer the whole tag set towards being a high quality index of our actual clusters of discussion. Theoretically. We're not there yet because it doesn't have any information to go on and hasn't been tuned up. The linguistic analysis stuff should give it that information. I'll be working on that.
The other main sofie dev wrote us a better tag selection dropdown that has simple search functionality. Check it out. It's easier than the old one to find the tags you want. But the real magic will come when we integrate it with the autotagger and new tag machine to suggest tags based on post content and other suggested tags. We're preparing for the problem of managing many dozens or even hundreds of tags, for which intelligent suggestions will be crucial. Check it out and let us know if you find any bugs.
This week intern-kun did some backend work, writing us a database anonymization script so that the whole dev team can work with real(er) data when testing. Don't worry, we're keeping your secret identities safe. He also worked on the newsletter stuff.
That's all for now. Let us know what you like and what you would like to see.
It's a pretty simple deal: the top five threads from the last two weeks minus anything that's already been featured in a newsletter. So if you post on Saturday and it doesn't blow up until Wednesday, you'll still be featured in next Monday's newsletter as a late bloomer. Basically, it should just work.
Hopefully it all works smoothly. We've tested it a bit, but as you might imagine it's hard to realistically test a periodic newsletter. There may be delivery issues, rendering issues, etc. Please dig it out of your spam box and tell your mail provider it's not spam. Report any bugs in this thread or to me by whatever means. We'll work out the kinks and get it ready for primetime over the next few weeks.
We are always happy to hear your feedback about any aspect of the system, especially at these early stages. Let us know how we can make this newsletter great for you.
Meanwhile in other progress, I completely rewrote the tag system backend. I hope you don't notice. It is now in a form where where it will be much easier to integrate more advanced features like autotagging based on linguistic analysis. I've got a plan and I think it's going to work. The autotagger will not only extrapolate from our tag use patterns to suggest tags based on post content, it will be capable of refining the boundaries of these tags and even autonomously discovering and managing new ones (which we can then rename and use as we see fit). It will generally defer to us, but will attempt to steer the whole tag set towards being a high quality index of our actual clusters of discussion. Theoretically. We're not there yet because it doesn't have any information to go on and hasn't been tuned up. The linguistic analysis stuff should give it that information. I'll be working on that.
The other main sofie dev wrote us a better tag selection dropdown that has simple search functionality. Check it out. It's easier than the old one to find the tags you want. But the real magic will come when we integrate it with the autotagger and new tag machine to suggest tags based on post content and other suggested tags. We're preparing for the problem of managing many dozens or even hundreds of tags, for which intelligent suggestions will be crucial. Check it out and let us know if you find any bugs.
This week intern-kun did some backend work, writing us a database anonymization script so that the whole dev team can work with real(er) data when testing. Don't worry, we're keeping your secret identities safe. He also worked on the newsletter stuff.
That's all for now. Let us know what you like and what you would like to see.
Hey anons, this week
