If they were saying to do an AMP version, I would agree, but it’s not the case.
They’re pointing to reasonable stuff:
the initial HTML takes 3-4 seconds (vs API query that takes up to 300ms)
They gave tips to reduce this part and indicated the React calls. Caching the rendered page html and only re-rendering user profile stuff could speed things up, or the renderToStaticNodeStream which seems to do something similar.
The caching can be done with NGINX reverse-proxy cache, or caching generated html in the server, or generating static pages and filling missing info (mostly user profile and interactions) with javascript like in Gatsby/JAMstack.
after that, loading CSS/JS take another 1-2 seconds
Could be loaded faster with an empty HTML that fetches the content instead of building the final HTML on the server (this seems to be exactly what the renderToNodeStream and ReactDOM.hydrate() does).
CSS and JS could also be minified to reduce size and parsing time.
I only understood about half of that, but I think I agree with you. If Google tells us ways that Musicbrainz is suboptimal in general, those are worth fixing.
But I’m grumpy about the idea (if it exists) that we would change our website just to benefit Google, or to get better Google-search results (and I strongly suspect that Google search rankings have more to do with what benefits Google than what benefits users)
Contrarian that I am, I want Musicbrainz to dig in their (/our) heels and say something like “dammit! We’re not going to let Google dictate how we design our webpages!”
But I’m grumpy about the idea (if it exists) that we would change our website just to benefit Google, or to get better Google-search results
Okay so, most of the rules/guidelines are not really for the sake of SEO, most are just usability, stability and performance. They’re holistically more good for users than Google, it is not AMP with it’s hidden downsides.
For example when using something like Lighthouse to scan for issues, SEO is a separate category you could totally ignore - but in my experience, those are relatively trivial to fix when you’re fixing other things. Plus, let’s not forget, making it easier for search engines is making it easier for the users to find content as well - how often do you visit sites you can’t find?
I have had a quick look at the metadata being included in the MB web pages, and I am not surprised that it does so poorly in search results as the title and description tags are not ideal, whilst other tags like keywords, character set and language are missing entirely. I don’t think the changes needed to improve things would be that difficult.
However, I believe that MB is on a par with Wikipedia in providing quality information, and if you do a search for an artist or album, I would like (or even expect) a box at the top of the right hand column (like you often get from Wikipedia) based on MB data. I have no idea how you persuade Google to put MB data in that position, but once we get the page metadata right it might be worth approaching Google to see if there is some synergy to be exploited.
P.S. Is it just me who sees a certain humour in the MetaBrainz Foundation - whose primary purpose is creating metadata (on music, books etc) - not being up to speed with delivering web-page metadata for google (and other search engines) to consume?
That box already uses MusicBrainz metadata (which is sometimes annoying, see this thread, but that’s a separate problem).
We didn’t pay a lot of attention to this historically because we used to want to avoid getting traffic from Google - in fact, we used to block indexing fully for many years. This was mostly due to a (hopefully baseless?) fear of an influx of new users overwhelming the site without taking the time to learn the guidelines and decreasing the quality of the data. That said, you’re right that if we’re going to be indexed we might as well do it as well as we can.
Having contributions that aren’t even indexed after months is also a problem for the motivation of editors: sometimes you feel like you’re writing in a black hole.
Sadly we have very little insight as to what Google does with the data they retrieve from us. We do know that they use some 40% of our traffic on the website portion of MusicBrainz every day (1-4 requests per second). We do know that this data goes into their knowledge graph (for which we get paid support!), but we have no idea how much of this traffic ends up in their search index. As you might imagine the search team is quite well insulated from the stark raving lunatic unwashed masses of the internet. We have some access, but asking them questions like “why do we not appear in search results” is pretty much assured to not get a response.
The fact that we have lots of stats, but little text on your pages and that our pages are not very mobile friendsly also doesn’t help us in the rankings.
That said, the longer term plan includes making MB more mobile friendly and to build out ListenBrainz and make it appeal to a broader segment of the music loving population. And we’re finally seeing some progress on this front – LB has more users coming in and more contributors as well. We’ll soon identify data that is missing in MB and prompt users to add this data to MB, which should drive traffic and eyeballs.
This worked quite well for us in the early days when last.fm recommended that people tag their music collections with MusicBrainz – our community grew substantially because of this and I expect that this will happen again as LB becomes a bit more used. While this may not help with our rankings in Google, it should help with the case that @PierPiero mentioned – languishing edits.
The weird thing is that this all applies to Discogs as well, but they are consistently in the top, for image searches and for artist/album searches.
I started using MusicBrainz (again) back in the day because of Last.fm! Not to tag my collection, but because they took their tracklists from the DB, so fixing info had to be done in MB.
I might be wrong but I think that currently and historically the only source of viewers and editors is via other applications, people using taggers, tools like Plex, maybe a few others… maybe LB in future. But is LB turning up on heaps of search results?
I think this needs to be dug into. MB relying on other applications to bring in the occasional editor is a really roundabout way to keep getting traffic/users. If we can identify something that fixes whatever blocks MB from turning up in searches, like putting (collapsible?) chat boxes on entities to get unique text/info, then it should be a priority. Opinions may differ…
This seems kind of like a pipe dream, but a unique shoutbox-style element for each MusicBrainz entity page would rock so much. Sure, there are already the forums, and people can write edit notes on individual edits, but if there were a spot on a page to just discuss an entity in a less committal way than making a forum account or signing up to post edit notes… that would be pretty dang rockin’.
To go deeper in this analysis, Discogs is now the reference site for music on phisical media due to their marketplace, but is starting to lose its grip on digital media release versus Spotify / Apple / Deezer databases. It is to be expected that it will soon become incomplete as AllMusic is today.
I fully agree with this strategy, hoping to start having soon some features available on this platform, such as the now missing Recommendation Playlists.
Integrating underutilized MB features into LB could help here. For example, giving genre breakdowns of your listening history would encourage voting for genres on MB, and search engines might point somebody looking for “[country name] [genre name] albums” towards MB.
Given releases that are both in Discogs and Spotify/Apple/Deezer (so with a similar amount of information on the page), Discogs performs still better in terms of page ranking in search engine results.
Given releases that are on digital media only, Spotify/Apple/Deezer perform better in terms of ranking than others like Bandcamp, even if the second one has usually more information on the page than the three.
So it’s a matter of search engine optimization and potential issues I would verify for MB comparing it to Discogs/Spotify are:
internal links: as discussed already in the new homepage proposal, the paths to reach the pages of the site are useful for users and search engine bots. The kind of links can range from simple alphabetical lists of artist, like in Jazz Archives, to rich categories of genres, like in Rate Your Music. In case of RYM, it performs better than MB even though it has less information about releases, with the list of recordings generally missing;
grouping information on a single page instead of splitting in multiple places: in Discogs you have pictures, links to related artists/groups and listing of releases on the same page, but in MB such information are scattered across different pages (cover arts, relationships, …). It’s not a case that if you are looking for an album, in case you get a MB result, it’s more likely to get the cover page rather than the overview.
I was pleasantly surprised (but only because my expectations are so low) that a google search for “bonzo dog band discography” turned up the MB artist page on only the second page of results. Unfortunately the same is not true when I don’t include the keyword “discography”.
i notice google doesn’t link to musicbrainz, but they still show our data, which kinda sucks… right now, we’re one of the only databases with taylor swift’s upcoming rerecorded albums and you can tell they’re using our data because all the tracks follow our eti guidelines and they have only the tracks we have in the database. i love the fact they’re using mb, but i wish we’d at least get credit… https://www.google.com/search?q=1989+taylor’s+version+tracklist
It’s good that people get all this data (I wish that were the case for movies and shows as well), but I don’t get why not index the source page or display a MBID. I get that we don’t want a flood of abuse, but some visibility would be nice.
Just searched both duckduckgo and bing for
bragod musicbrainz
No musicbrainz db return. (Though a mention on this forum gets found by both.)
Seeing Bragod’s “official site” and Wikipedia page being returned and highlighted by Bing though.
Suspect “official site” link is being taken from WP.
Wondering if getting MB page put into WP “External links” section alongside “official site” would get MB some search engine traction.
Or maybe MB just isn’t ready to move away from being a niche project into the mainstream. This possibility needs to be considered carefully - being suddenly swamped with both volume and vandals would have real dangers. Maybe just having techno-elites as the userbase is better for now?
