Evaluating sites

For those who pay attention to certain corners of the internet, December brought another major upheaval. Beginning on December 17th, Tumblr made a major turn in how they intended to handle certain kinds of content (and among other issues, applied a computer-driven algorithm to figure out what was banned and what wasn’t. Dear Internet, our AI is not actually that good yet.)

This means it’s time for another in the periodic posts I am inclined to make about how to learn about resources.

I have been around the Internet for approaching 25 years (basically, since I got to college in the fall of 1994, so in 9 months or so, I will have a party.) It was the dawn of time, when having webpage backgrounds some colour other than pale gray was new and exciting. Since then, internet sites have risen and fallen, come and gone.

It makes a person (well, a person like me) cautious about who has that information and what they want to do with it.

A word about my experience: I’ve been on various sites as an active participant since 1999 or so (and Usenet and mailing lists before that). My primary online homes have been LiveJournal, Dreamwidth (the current), an online Pagan forum (also current), and I dabble in Twitter, Tumblr, Facebook, and various others as relevant . I’ve been staff on the Pagan forum for a couple of years now, and I spent 18 months (in 2003 and the first half of 2004) as a volunteer on the LiveJournal Terms of Service Team, back when online sites were still figuring out a lot about the basic categories of what users will try and do on online sites (both in the positive sense, and in the ‘what can we get away with’ sense.)

Who is the customer?

One of the first things to know about a site is who the customer is. On sites with advertising, that’s not you. That’s the people who want to sell ads. This brings a whole set of choices to a site, many of which are complex to negotiate, and often don’t come off in favour of the people who are producing content. In fact they don’t want to show ads to you – they want you to create stuff that makes people read your stuff, so they can show ads to your readers. And learn about your readers, so they can show them more ads.

Years ago, Denise Paolucci wrote a series of three posts about why monetizing social media content through advertising was doomed to failure. (Part onepart two, and part three). I thought she was right then, and I still think she’s right.

(Disclaimer and context: those three posts were made while Dreamwidth, a project and company she co-founded, were in beta development but before it was open for more than a handful of test accounts.

Dreamwidth makes its operating costs from paid accounts which come with benefits from a small percentage of users – most accounts on the service are free, and they built the budget assuming that. To be utterly clear, I give them money every six months for a premium account, and not just because I believe in feeding D’s cats.

Basically, if the site is built for investors, or using venture capital, there is a very very good chance that you (the individual user) are not the customer, you are the product. You may decide you’re okay with that, but I believe it’s important to understand what that means. And most sites in this category go out of their way not to tell you.


What kind of experience do the people running the site have? A lot of people who start sites have a surprisingly limited experience of how other people use the tools they’re using (and therefore might want to use the tools they’re creating…)

There are lot of things that come up for sites that have user-created content (as opposed to say, articles that go through editors and an editorial/publication process first. End users on sites will come up with all sorts of ways to use them, many of which the people designing the site may not have anticipated. There are a few basic categories that seem to be constant, though:

The internet is for cat pictures: roughly speaking, the longer a given site is around, the more people will want to use it for pictures of their cats, their dogs, their kids, and various other animals and beings in their life. Some of these have more privacy issues than others. A well-planned site will be thinking about how to handle this.

(Also, if you have not already read Naomi Kritzer’s award-winning Cat Pictures, Please, I recommend it.)

The internet is for porn: A special category of ‘beings’, these uses obviously bring a lot of legal issues with them, both in terms of what is legal to display, to which users, in which countries, and in terms of whether it’s okay to post things on that site.

Defining porn is notoriously difficult: how do you create a policy that understands the difference between porn, breastfeeding, classical and fine arts, and different kinds of bodies? The current AI, for example, seems to be rather biased about skin colour, because of the material used for training. What’s the difference between drawn materials and photographs or videos? What’s different about text? (The law handles these things differently, so site policy probably needs to take all of them into account, unless a site is text only.)

The trick is that people are endlessly creative in this area, so a site needs to have a good way to think through the issues (and what’s allowed by their hosts, payment processors, and other external forces) and how they’re going to communicate that to their users and enforce whatever rules are in place.

The internet contains people. Not all of them are well-meaning. A site that isn’t thinking about the issues of harassment, stalking, privacy violations, etc. is probably not a site you want to be around much.

A thoughtful site policy will need to figure out what kinds of privacy violations they act on (is posting a legal name, address, or contact information a violation? What if it’s truly public information, widely available online? Is it okay if someone posts their own information, but not okay if there’s incitement to harass someone?) How do you give users control over who sees their comments, who can access their space or comment in it? What do you do about accounts designed to impersonate or harass someone?)

Sites make all sorts of choices about what tools and options they provide – and these lead to different options about how to deal with harassment, or even just irritation.

Periodically the Internet blows up. If a given site is setting itself up as an alternative to a particular option (filling a specific kind of niche or type of use), then they should be planning ahead for what happens if the dominant player(s) in that part of the Internet have highly public issues.

Can they take on new users? Can new users figure out how to use them? Will the servers hold up? Sites that aren’t thinking about this (even if the game plan is “We are staggering new users this way, here’s how that works…”) are going to struggle with a lot of other things.

What kind of financial planning has happened?

And how transparent is the site? Some sites start as a passion, built by one person or a few people, then slowly adding more people. They may rely heavily on volunteers (this has some complexity, but can work really well if the site understands how to support and train volunteers). Other sites start as an idea, get funded by venture capital, and can add staff quickly. Others are somewhere in between.

Total transparency isn’t necessary (or realistic) but you want a sense of how many people are working on the project, what happens if some key person isn’t available for an extended period, etc.

Here’s the thing: programmer salaries in the United States are expensive. So is health insurance. Sites often need other employees than programmers (sooner than later, anyway.) Servers and bandwidth are cheaper than they used to be, but running a large site still has significant ongoing expenses. A site that doesn’t have a good plan for meaningful ongoing funding is going to have problems.

(There are a lot of different ways to handle this, but a lot of failed sites have seriously underpriced a lifetime membership, or assumed that advertising will make up the rest, or – well, there are a bunch of failure modes. There are also some long-term successful options, and they don’t necessarily have to cost a lot of money.

Metafilter, for example, runs on a small fee for account registration, and then asking people to donate as well as some advertising shown to non-members. Dreamwidth assumes that a small percentage of people will have paid accounts, and based their pricing on that assumption, with extra fees for some fun but not essential tools like more icons.

If you can’t figure out the funding plans, that’s probably coming from you, and it’s probably based on other people benefiting from your content, possibly in ways you will not like in the long run.

What are you committing to?

There are two big risks with a site that’s badly run, and they both come down to the same thing: can you get your data and content (including connections to other people) out easily, on short notice, and without needing their services to be robust?

Sometimes badly run sites will make a policy change (usually on short notice at a horrible time for you) and you will need to decide what you’re doing about that with not a lot of time.)

That’s a really bad time to realise you have a lot of information (your writing, your art, your comments, your journals, your connections with other people) you can’t get out. It’s a good habit to figure out backup options anyway, but it’s especially important with a site that might make sudden changes or hit financial difficulties without warning. Such as sites that have been sold to new owners, or just hit prominence in the news.

(That’s the other reason to be cautious: a site having financial trouble may announce a planned shutdown, but sometimes things just disappear.)

A well-run site (that understands how the history of sites have gone) will give you ways to export your content, or make it clear what can be saved and how. They’ll be up front about portability or cross-posting or other tools (you may have to look at help information: sites with lots of features can feel overwhelming). And they’ll communicate with you when there are problems about what’s going on, in some form other than a brief error message or a purely-PR-speak statement.

Bookmark the permalink.

Comments are closed.