1

Specifically, Student5266’s article:

https://www.wikihow.com/Upgrade-the-RAM-in-a-Lenovo-Ideapad-S510p-Laptop

It has come to my attention a recent post undoing an edit that reverted the wikiVisual images to the photos that the author himself took. Here is a revision on what I believe should be the current status of the article:

https://www.wikihow.com/index.php?title=Upgrade-the-RAM-in-a-Lenovo-Ideapad-S510p-Laptop&diff=21510868&oldid=19837336

The original photos that the author took are good. They are a bit blurry, they are not great or HD as we would like them to be, but they do the job because they are real images taken by someone who has the specific model of the laptop. The red marker highlights specifically where screws should be loosened and how the whole thing is supposed to look. The images clearly present substeps in a reasonable fashion, such as removing the battery. The electronics are detailed and the components are tiny - yet another advantage of taking *real* photos. If I wanted to improve the article, I would probably take some higher quality images. Again, they are not the prettiest thing in the world, but they do the job.

Contrast that to wV’s second attempt in March of 2016 (the first attempt back in August of 2015, to put it lightly, not the greatest thing), which did a better job than the first time around…but these images are still cartoons, that are vague and unhelpful for readers who are not familiar with computers or who are not entirely comfortable with messing around with a laptop’s internal components. They are neither high quality real images that should represent an effort by a project that has been around for quite a long time, nor are they a substitute for that particular model, for an article with that specific of a title. They miss the substeps that should be done and happen to be important for the task; I hope readers would know to remove the battery before doing anything with the electronics, but again, an image that clearly conveys that is just better for the article’s presentation. I personally disagree with the overall aesthetic and look of wV, but that’s another discussion. All of these factors point towards a reversion back to the author’s original images.

This is another example of an effort that has gotten past RCP without a second thought, I would think much to the chagrin of the original author who has made some effort to revert these attempts, and even if he is happy with the new images, there are plenty of other authors out there who have complained about their well-taken photos that keep getting replaced by vague, cartoon drawings. These replacements are sometimes good whenever an article really needs step-by-step photos, but:

a) they are not always needed, 

b) there are still too many instances where they have hurt the integrity of the article by presenting visuals in a more unclear, vague fashion than before, 

c) their edits are subconsciously viewed as more authentic or “good” than a similar hypothetical effort by a volunteer, patrolled without a second thought to their actual quality,

d) they are occasionally silly or of such low quality to the point where other websites and games mock these images and hurts the brand and reputation of this site (e.g.  https://www.wikihow.com/index.php?title=Cope-With-a-Double-Parachute-Failure&diff=21024996&oldid=21024906 ), 

e) they have been making mistakes at a rate that no reasonable person should be making (c.f. math articles - is it really excusable to still be missing superscripts? Inverse power is *not* the same as minus 1), and

f) they have (courtesy past forum posts that brought up the issue about Seymour Edits) even presented advice that has been outright dangerous at times. 

I have suggested that some categories like math should be held off from wV because of the exhorbant amount of errors that they produce, or a category like chess, where I have suggested that a 2D board would be better for a reader than a 3D board, which is harder to see on a screen. The latter, by the way, should be common sense, even for a person who has never played chess before.

I think we need to scrutinize these efforts, to not be afraid to revert just because it is wikiHow-sanctioned, and to understand that they are not necessarily the work of people who have actually done the task or improved it because they wanted to. 

2

I saw this issue pop up yesterday and reached out to the author (as had some other editors, too). I wanted to touch base with them before deciding what to do next, mostly to ask if they had any questions about the wiki (they seemed surprised others could change the article) and to follow up about the possibilities for next steps. I was waiting to hear back about their thoughts on a revision with a combination of images, but it looks like they haven’t had a chance to reply yet, so I’ll just go ahead and restore their versions for now, and if they like the idea of a combo or redo, we can work that out later. 

There’s no problem from the wikiHaus POV with doing a revert or a combination edit or a redo here. In an ideal world, like you said, the images would be clearer - either as less blurry/more uniformly sized photographs or more precise technical illustrations. But absent the ideal, we don’t have a problem going back to the originals here, even if the author doesn’t reply with their own preference. And the visual team definitely has no problem with the folks who know topics like this well undoing images, asking for revisions or offering feedback when added visuals don’t do the steps justice. The scrutiny is welcome. This is actually a good time for this example to come up, too, because Connor has been working with several illustrators on new experimental styles for technical illustration: approaches that are less cartoony but still illustrated in ways that help visually highlight the important aspects of the step (which can be harder to distinguish in some photographs, depending on the topic and background). I don’t think they’ve experimented with that style on any DIY electronics topics yet but maybe it’s an area to try out and get reader feedback on (not necessarily with this article, since the photographs probably do the trick here, but in general). Thanks for the food for thought on this one!

3

I’ve noticed it as well where once (a while ago) I had taken pictures for a specific article that wikihow had already added photographs to, but didn’t follow the article, so I took new pictures myself, but wikiVisual ended up reverting them when I knew exactly what the steps were doing. I ended up reverting wikiVisual.

It’s a good idea to check in RCP which pictures are better and not just to click “patrolled”.

4

Just a note to everyone, I told Atheia to start a discussion here because there were other editors who did not agree with his point of view.

While I agree that the original photos should maybe be replaced with more professional photos that are a bit clearer, the illustrations were not a step in the right direction. There were details lost in the illustrations from the original photos like screw placement.

5

Wikivisual automatically patrols their own edits. Their edits never come up in RCP unless it’s the rare occurrence of an edit getting bundled in with the upload. I don’t see why the patrollers are seemingly getting blamed for something they don’t even patrol…

6

I don’t think patrolling or autopatrolling is the main concern here - these aren’t really the kinds of things that are likely to be caught in patrol even if bots went through there. The best judges of what images work or don’t work are often the actual readers using the article, and while this article doesn’t have a high enough readership to get a ton of helpfulness votes, the small amount of feedback gathered on this one does suggest that the previous version was more helpful. We have no problem undoing a case like this (I have done so), and are happy to redo/adjust/fix/etc when someone who knows about the topic spots a problem, or when reader helpfulness/feedback suggests there’s an issue.

@KommaH ’s point on screw placement is actually the kind of thing I was thinking about in mentioning the experiments with technical illustration. Being able to get more precise about that kind of detail while avoiding any issues with clarity/blurriness/background/darkness of a photo, etc etc. That could be a win-win. You can see the kind of thing I mean on  https://www.wikihow.com/Tie-a-Tie - Personally, I kind of like this aesthetic of keeping the background simplified/plain but highlighting the important bits. It’s early days for that kind of style, so we’ll see where it goes and what readers think of it in the long run, and I’m not saying it would work in exactly that way on this topic. But I do think it holds some cool promise. It could be especially neat since so many people browse wikiHow on mobile devices now, and the smaller screen can make it hard to pick out details from a photo that can be made extra clear in a bright illustration. 

7

It seems like wV not going through RCP is a very recent addition - just a month ago or so, I still saw their edits while patrolling. If this is the case, then it just makes it even more egregious that they would be allowed to bypass the normal rules. It is not just that their edits are viewed as “better” than similar work by volunteers; their edits are allowed to bypass the first line of defense.

This is unacceptable. A patroller may not be an expert in everything, but I am confident that he or she can use their common sense and figure out if adding images would be more helpful or not in the majority of cases, especially if there were images taken by volunteers in the past. If you can tell, then patrol or revert; otherwise, just skip it. This simple idea applies to any edit seen in RCP, not just wV. Part of the point of writing this long post was to call for patrollers to be more vigilant in patrolling wV edits. If it is the readers who are the better judges, then this is simply a slippery slope that leads to the question: what is the point of patrolling? To simply revert spam and vandalism? Part of being a good patroller is to judge whether an edit improves an article.

In this case, given such a specific title, the most logical step would simply be for someone who has the actual laptop model to take some higher quality photos - otherwise, forget about it. Why is there a need to hire a bunch of artists who probably don’t have much experience with working with computers anyway? If such cartoons end up being more and more realistic, I don’t see a reason why real photos shouldn’t be taken at that point. And if such an effort is too expensive or otherwise too difficult, then that’s their problem of judgment in thinking that they can produce something better than the original author, who again, has actually done the task.

Furthermore, none of my original points have been responded to in any way. Neither my 6 points I outlined, nor my two suggestions for two categories below them. Since wV seems to be allowed to bypass RCP nowadays, these complaints should be viewed as even more urgent.

8

The Wikivisual bot not going through RCP isn’t new. Some manual image additions go into RCP - those are the “library” images that are manually added in wikitext from the existing image collection, through the WikiHow Projects account. But the bot uploads of new images don’t go into RCP, much like most bot edits don’t go into RCP, and that isn’t new. Hopefully with some background context, the difference will make sense:slight_smile:

Patrolling is a great way to get a second set of eyes on an edit to make sure it’s not spammy/vandalism or doesn’t have errors in format and so on. But patrolling isn’t perfect either; we as patrollers are not expert on all topics (if we required all volunteers patrollers to be certain about accuracy on every edit they patrolled, we would never get anything reviewed!). There are plenty of times when I’m not sure about a change I see but assume good faith and let it stand, trusting that the wiki methodology will work. Wikis aren’t perfect - but if I approve an edit even if I’m not 100% sure of its accuracy, those errors can be caught and fixed, because we’re a wiki. The kinds of issues patrolling is good for catching are more of a concern for the wikiHow Projects additions, since they’re added manually and might come with some human formatting errors every now and then. In the case of bot uploads, we don’t have to worry about some of those issues, since layout is taken care of by the bot.  

With all changes around the wiki, we rely a great deal on reader helpfulness ratings and feedback to tell us what’s working. Those readers are the actual folks using the pages and trying to follow along, and our joint mission as a community and staff is to serve them, so where patrolling can’t catch issues, it’s good to know that the reader experience data is there to guide us. With something like bot uploads, patrolling seems like a waste of energy, especially when there are many, many image additions, and the RCP tool is not catered towards image review (all the patroller would see in the main diff would be image tags). Jack has always felt very strongly about wanting to avoid creating unnecessary work for volunteers in that area, which is why these additions don’t and haven’t before gone into RCP; this is similar to how admins can autopatrol, to help avoid cluttering up the RC list and creating unnecessary review work in cases where we already know the person/project is contributing in good faith. 

When it comes to your other points on image style, the visual folks have done and continue to do a lot of purposeful experimenting with reader preferences. They try different approaches, in particular in different topic areas on pages with high readership, to see how readers find them - and then trickle their learnings down to other articles too (which may not have the readership to run fruitful experiments on directly). This is what they’re doing with the more technical illustrations now, for example, so what they learn on those topics will likely affect what they do on pages like this in the future. This kind of experimentation is admittedly tougher to do in areas like the niche math topics you work in most often. We totally understand - and share - your frustrations when there are errors in the visuals there. Our visual folks are not experts in all of the areas they create images for, just like most wikiHow community members are not experts in everything they edit - so when mistakes do happen and haven’t been caught in review, we appreciate hearing about it for follow-up. We are also happy to exclude articles from image additions, and have done so with the articles you’ve started and the ones you’ve let us know you’ve done big expansions on. If there are others, let me know! If Student5266 would like this one excluded, too, we’d be happy to do that on this topic as well. 

I will say, though, that the overall approach of adding step-by-step images/illustrations to articles is generally very well received by readers. It has been, in many ways, a game changer for wikiHow over the years. People often identify and remember us by our images, and much more often than not, they find them helpful. We do certainly want to fix silly or inaccurate ones, but we also do believe strongly that the overall visual project, even with its imperfections, is helping us to fulfill our mission. It’s not perfect, but for every image set with an issue, there are many successful and helpful sets. We have a lot of happy campers as a result of their work - both community members who request images they can’t make themselves and appreciate that the visual team will tackle it for them, and readers who appreciate having another medium on the page to learn from.

We know the project isn’t perfect, and don’t hold our breath that it ever will be - we’d love it if we could achieve that, but Jack’s guiding light for everyone contributing to wikiHow has always been that the wish to be perfect (and inability to be, since we’re human) shouldn’t hold people or projects back from doing things that work towards our long-term mission. Staff efforts will never be perfect, just like volunteer ones won’t; we agree that these visual edits shouldn’t be seen as any better than any other contributions. We are happy to fix errors or inaccuracies where issues come up. Please do keep sending your feedback when you have concerns, because that helps us correct mistakes, weed out unsuccessful approaches and illustrators, and improve the project more quickly than we typically can from reader feedback alone.  We may have to agree to disagree on the exact approaches the visual team takes in some cases, since I know you’re not a fan of step-by-step images on many topics, while our reader data suggests they, more often than not, find it helpful. But we are all on the same side of trying to achieve the mission, even when our exact approaches aren’t identical.