I often see people recommend that you only play from “Urtext” editions of music (as opposed to “Edited” editions), because these editions are designed to show only the composer’s “true intentions” without the (probably erroneous and apparently often distasteful) suggestions of an editor. This advice is not necessarily bad, but it needs additional context and nuance in order to actually be useful to you.

First, you have to know what an Urtext edition actually is. This is an edited edition of sheet music (yes, even Urtext editions are edited), but edited by scholars who make it their job to go through as many authoritative sources as close to the composer as possible to try to put together an edition that reflects the notation the composer seems to have intended in the end. This point is critical: it’s the notation the composer intended, not the performance the composer intended. One of the big dangers of using an Urtext edition today is to read it like it’s a modern piece of sheet music - if you do that, you will probably play the piece wrong, or at least not the way the composer intended (we’re assuming here we’re talking about a historical piece from the baroque, classical, or romantic periods, or maybe even impressionistic / early 20th century). It is only really in the 20th century that composers start trying to notate as much of the performance as possible, before this period more of the musical interpretation is left to the performer. 

This is both because there was more often an assumption that there is “one” way to play music (you’ll often see treatises with names like “On The True Art of Playing Keyboard Instruments” (C.P.E. Bach), as if there is only one correct way to play music), and also because when you only have one person playing solo keyboard music for example, there’s no need to coordinate across multiple players for interpretation, so leaving more to the player is nice for flexibility and judgment, and making the player feel like they are involved in making the music, and not just mechanically executing something from another person. This was also an era where in this (classical) world people largely wanted new music and so the composer was frequently involved in the preparation and performance (they either were themselves the performer or one of the performers, the conductor, or a teacher working with their students). So they didn’t need to be as explicit in the notation, because they could explain what they wanted if it wasn’t working by default.

One example of where this comes up in practice is in the keyboard music of J.S. Bach. I’ve seen people try to claim that you shouldn’t use expressive dynamics in Bach’s music because if you look at the Urtext editions, they will show no dynamics (or at least very few dynamics) in the notation. This is bad advice, because Bach didn’t actually intend for there to be no expression in his music. Little of his keyboard music was published in his lifetime - it was mostly used as teaching music for his students or shorthand notation for himself to play. He would know what sort of expression he might want to add, and so there was no need to write it out if it wasn’t going to be played by people that he wasn’t directly working with very much. But even more than that, the other reason not to include a lot of expression was because in that period (the high baroque), there were really three keyboard instruments in common use in his region (none of which was the fortepiano, which we now call the piano in its modern iteration).

J.S. Bach mostly played harpsichord and pipe organ, and also clavichord. These three instruments are related mostly in the sense that they are played from keyboards, otherwise they work quite differently. So you couldn’t write one set of expression marks that would be usable across all three - they could play similar music, but not similar expression (or not the same way), so there’s less of a point in writing in expression as that either limits the utility of the music, or the players would have to ignore expression on a different instrument and do something else anyway. But even the harpsichord can be a very expressive instrument if you play it as a harpsichord and not as a weird piano that doesn’t have velocity sensitivity for dynamics. So these are reasons why there aren’t expression marks in Bach’s music. 

It’s actually similar to modern general tunebooks in fiddle music. In the world of Irish Traditional Music, for example, there are a variety of instruments that can be used to play the music - fiddle, flute, tin whistle, bagpipes, various kinds of accordion, and nowadays mandolin, tenor banjo and guitar among others. All of these instruments can achieve similar results in expression, but not in the same way (mostly here talking about phrasing and ornaments). So the generic tunebooks will be written with just the notes on the page, no slurs / phrase marks, ornaments, or dynamics. It’s expected that you learn to play your instrument(s) and how to add expression on them the way each one does itself, and then you add that in yourself. If you try to read a tune like that as if it was a modern classical piece, it will sound completely wrong to anyone who knows the tradition.

Someone once pointed out that Haydn (or maybe Handel) had a piece that he wrote for solo keyboard, and then scored for string orchestra, and in the string version there are many dynamics, but not in the keyboard version. This again is not because he expected an expressionless keyboard performance, but because there’s no coordination problem like there is with a string section (all the players have to agree on the same expression within each section, and it’s best if there can be a master plan across the whole group too). So solo keyboard players can add the expression they want based on their instrument, and therefore there’s no need to lay it out for them - back then that might have even been insulting, as if you didn’t trust your player to have decent taste in music.

The other reason why they could get away with this back in the baroque and classical periods in particular is because this was the heart of the “common practice” period, so most people who would have played this music were trained in a similar fashion, and composers wrote in a common practice and therefore as the player you could assume that the way you were trained to play was the way the composer would expect you to play the music. Later in the romantic period and especially in the mid-to-late twentieth century, things started to fracture and composers had different ideas both of how to play music and how to notate it, such that it’s less safe to assume the way in which someone will interpret your vague notation and you should be more explicit. This is also the period when composers and performers start to split more - in the baroque and classical periods composers usually were performers and performers were able to compose and improvise as well. As the nineteenth century progressed, and particularly as universities became primary patrons of the classical arts, the roles became more distinct and composers started writing more for other people to play, so that they needed to be more explicit. This starts happening more with Beethoven, who was one of the first composers to notate more explicitly (though even he leaves a lot to the player’s judgment).

So where does this leave us today? I think some modern composers over-notate their music now, trying to specify every detail and leave nothing to the interpretation of the performer. I’m not sure it’s actually fundamentally possible to do that completely, but some people try. I think it has to do with both a twentieth century idea that sheet music is the real music and the sonic realization is a flawed version of the sheet music ideal (this gets music precisely backward in my opinion - see General Musicianship Tip No. 2!), and also the idea of the performer being the mechanism through which the composer’s vision is realized, but with the composer’s intention being paramount. This also is at least a little problematic - see Composition tips nos. 21-22. Note that I’m not trying to argue that performers shouldn’t try to realize the composer’s intentions, but that treating performers as if they are just machines being played by composers with no musical will or sensibilities of their own seems problematic to me. But nonetheless, our modern approach to sheet music tends to aim for marking any significant changes in the music, such that if there is no indication of change, then you need to keep “natural phrasing” subtle. Some people try to argue that if it’s not marked, then you should do none, but I think that might go too far. This may differ from composer to composer today, so if you can find out how the composer you are playing thought or thinks about this, you should defer to them. But for me personally as a composer, if I want no expression I’ll say so (senza espressione, non espressivo, or something to that effect). Otherwise, I expect natural phrasing and general musicality in the performance of my music even when it isn’t explicitly marked, and then to exaggerate that when I do mark something.

For the music of the past, Urtext editions are great either if you have a teacher who can help you interpret them correctly, or once you learn the performance practice of the period and composer in question sufficiently that you know how to play the notation as written in the Urtext edition. This goes for tunebooks also - they can be helpful once you have a thorough understanding of the nature of the (fiddle) music (including that the music is a living thing that shouldn’t be played the same way each time, and some idea of appropriate means of variation). I have personally found that I learn this music better by ear, either being taught by ear from workshops or ensembles I’m playing with (these days the Scottish Fiddlers of Los Angeles) or from recordings, but tunebooks or written versions of the music are a useful thing to have around, and I have a lot of those as well.

Before you have this understanding, or if you don’t have a teacher to help you, I find “hybrid” editions the most useful. That’s my word for them, not sure if other people use this. In the fiddle world, books aimed at your instrument with appropriate expressive markings added will be most useful (and probably necessary, along with listening to recordings or live performances by good players). I know solo keyboard editions better than other classical areas, and in this environment, my preferred editions are often the ones by Alfred Music, which I consider to be one of the best middle grounds between Urtext and fully edited. The way they do most of their editions is to show their “Urtext” rendering in solid black print, and then to add editor’s suggestions in either light gray or in parentheses, clearly marking when they’ve added something and when it’s from the composer. They also use a lot of footnotes to render things like ornaments, or to note when the sources they consulted had significant conflicts. The frontmatter in their editions is often very good too, and is useful to read particularly if you don’t have a teacher to help you learn to interpret things correctly. The benefit of this hybrid approach is that they can mark in notation to help you interpret the music better with modern notation practice, while still showing what the composer’s intended notation was as far as they can tell, so that you know what you “should” really pay attention to and what you might be able to tweak if you have an understanding of the music and want to do things differently from that editor.

I’d be careful about using free editions of the music on IMSLP or similar places. It’s nice that they are free, but many of the editions on those sites come from a period when editors felt free to “improve” the music of the past (often by assuming that their present performance practice was “better” than the past, and therefore trying to bring the music up to what it might have been like had it been written in that period instead of when it was), and they often do this without comment. IMSLP is a great resource, but look into more modern scholarly, Urtext, or hybrid editions as well. For solo keyboard in general, I'm partial to Henle and Barenreiter Urtext editions, and there are other good ones too, particularly for specific composers. If I've used modern editions for orchestra or chamber, they've usually been Barenreiter; though in that context it's often better to use the same edition as the rest of the ensemble if you can (and they will usually provide the music in orchestra).

The point of all of this is to create your own interpretation of a piece of music that reflects what the composer would have had in mind as well as your own personal taste, so you can make the music yours while being respectful of the people who wrote the pieces. But as you can see from this analysis, blindly using Urtext editions without any regard for notational practices of the period in question can be just as dangerous as using a badly edited edition, because the notational intentions of the composer may not reflect the precise performance intentions of the composer if read with a modern approach to notation in mind.