Metadata is one of those things that is generally seen as "well, I know I probably should, but, is it really worth the time...?". And the answer is, for most of us, at least the basic parts of it probably should be. There is a LOT of information that can be put into metadata, depending on what standard you're using. However, a lot of that may not be all that applicable to your particular datasets. What is likely applicable to almost all datasets though are a few basic core fields, especially when you're including the data for public release.
First, you should definitely put at least a basic description of what the data is. For example, the feature class may be called "County Roads", which seems self explanatory, but a description could briefly note something like, "This feature class contains the name, address information, and approximate location of all officially recorded publicly and privately maintained roads in XYZ County, TX." Describing the dataset allows you to explain to someone else what you meant the dataset to represent. For example, you may know "County Roads.shp" means all the roads in your county, but someone else may have thought that was a file of all of the county maintained roads in some undefined geographic extent.
Second, as a local government entity in Texas distributing geographic data representing property boundaries, you should be adding a disclaimer to your metadata as yet another way to ensure your legal obligations are met regarding having the state mandated disclaimer about "This is for informational purposes and may not have been ... ". This data could I guess go in a couple of spots probably depending on which metadata standard you're using, but in general there should be some sort of use limitations section where I generally put such a disclaimer.
Third, have you ever gotten a feature class from someone else and said something like, "What the heck is this %^&*@ ? They said they would send me the water-ways data for the county, not whatever this is." Or said, "Great, I now have parcel polygons but which of the following fields actually represents the total appraised property value 'value','Value2','OtherValue','PreValue','EVa'?" Well, metadata is your chance to prevent that from happening to someone else. If you asked for streams data, it may have been accurate to the person at a national office somewhere digitizing off a 1:24,000 scale map, but it may look completely worthless to you when you zoom in. If a dataset should only be trusted up to a specific zoom level, that is useful. Also, if you update the parcels you post for download only annually, the fact that the data shows the parcels as of the first of the year vs. up-to-date data may be relevant to someone, especially if they're downloading it nearly a year later and have no way of knowing if it was updated yesterday, a year ago, or the turn of the century. Or worse yet, think if someone posted a dataset "Historic Parcel Data for XYZ Area.shp", would you have any of knowing, without metadata, if "historic" meant 2 years ago or the 1950's? And then regarding other usability information like non-self-explanatory field names, metadata is the ideal place to make unusual field names more useful. If you have fields named Value, Value2, and OtherValue, you should document in the metadata that Value = Total Appraised Value, Value2 = land only value without improvements, and OtherValue = total improvement value before homestead exemption is applied.
Finally, even if you put nothing else in your metadata, you should put contact information. That way, if someone has a question about the data, especially if they get it passed on to them 2nd, 3rd, or 4th hand from someone else, they can at least look up who originally produced this feature class and can be contacted with questions about it.
Now, that all said, there could be much more that goes into metadata, especially if you were a federal agency under mandate to produce such per specific standards, wanting to participate in a data indexing repository such as a data.gov type site, a research firm wanting to wanting to formally document your data's origins and methodologies as part of a submission for XYZ permit or license, etc... But, at least including some basic metadata is definitely important.