3 ms·

Already now that site gives me significantly better results for certain queries (try searching for something along the lines of dual boot windows linux or

Already now that site gives me significantly better results for certain queries (try searching for something along the lines of dual boot windows linux or git fixup or similar on marginalia to see what I mean.) Warning: it seems to be a one person thing and my experience is that results might differ somewhat from week to week. Edit: someone usually point out that SEO spammers will arrive shortly. I say that if SEO spammers start optimizing their pages by making them useful, lightweight, readable and without most tracking and ads I see that as a win.

marginalia_nu 5y ago
The results are a bit hit and miss, I think it has to do with the fact that either there isn't content that match my criteria in some topics, or the spider is stuck in some local minimum and simply hasn't discovered it yet. > Warning: it seems to be a one person thing and my experience is that results might differ somewhat from week to week. Yeah, it's still very much in development and I only have one server so unfortunately I have to do some of the develoment and testing in a way that is very visible. I can usually do preliminary tests off another computer, but it doesn't have nearly the same resources so it can only deal with a sample of about 2 million documents, makes the test results pretty hard to interpret sometimes. > Someone usually point out that SEO spammers will arrive shortly. I say that if SEO spammers start optimizing their pages by making them useful, lightweight, readable and without most tracking and ads I see that as a win. I'd also like to add to this that since some of the things I use as quality measures are opposite to the quaility measures used by Google, that means that any site that attempts to optimize for both metrics underperforms in both search engines.
- eitland 5y ago
  Please be aware that already at its current level it beats its competitors hands down in not being annoying :-) Edit: > I think it has to do with the fact that either there isn't content that match my criteria in some topics, or the spider is stuck in some local minimum and simply hasn't discovered it yet. This is totally fine with me: Telling me the search engine doesn't know of any good matches for my search is not only a valid answer but also a far better answer than a list of a billion pages that some dumb AI somehow can construe as being related. In fact the list of non-matches has a negative value since I end up sifting through a hood bunch of them in a year hoping that this time there was something usable there.

marginalia_nu 5y ago

The results are a bit hit and miss, I think it has to do with the fact that either there isn't content that match my criteria in some topics, or the spider is stuck in some local minimum and simply hasn't discovered it yet. > Warning: it seems to be a one person thing and my experience is that results might differ somewhat from week to week. Yeah, it's still very much in development and I only have one server so unfortunately I have to do some of the develoment and testing in a way that is very visible. I can usually do preliminary tests off another computer, but it doesn't have nearly the same resources so it can only deal with a sample of about 2 million documents, makes the test results pretty hard to interpret sometimes. > Someone usually point out that SEO spammers will arrive shortly. I say that if SEO spammers start optimizing their pages by making them useful, lightweight, readable and without most tracking and ads I see that as a win. I'd also like to add to this that since some of the things I use as quality measures are opposite to the quaility measures used by Google, that means that any site that attempts to optimize for both metrics underperforms in both search engines.

eitland 5y ago
Please be aware that already at its current level it beats its competitors hands down in not being annoying :-) Edit: > I think it has to do with the fact that either there isn't content that match my criteria in some topics, or the spider is stuck in some local minimum and simply hasn't discovered it yet. This is totally fine with me: Telling me the search engine doesn't know of any good matches for my search is not only a valid answer but also a far better answer than a list of a billion pages that some dumb AI somehow can construe as being related. In fact the list of non-matches has a negative value since I end up sifting through a hood bunch of them in a year hoping that this time there was something usable there.