Latest updates on The Project

 Hello worldly people,

There has been a few updates to the bot these past few days, and I'm still chasing a not very important small bug, but a fairly annoying one. But first the updates.

Old Look, new look

If you were lucky enough to notice Urma's visit, you might have seen her old avi. She recently got improved, on the looks side of it. This was her looks at launch. I sat at launch  because she did have a few different looks over the years, mainly when I was experimenting with the skeleton code. Eventually, I ended up with this look:

 

And this is her new look:


I wish I could take the credit, but I've had a lot of help.. and we had a questionable start at best, so I think overall, she's great.

Additional Information

In addition to the gathered information described in the earlier post, we are now also gathering group info. Not membership, (I have no need to know who is in which group) but things such as group name, description, cost if any.  So if I was looking for a group that has, let's say, 'Blues' in it's description, I can do a quick search and tell you that I have found 8 so far, I can see the description, cost, number of members (not the members themselves, just the count) so I know how big of a group it is, in case that matters and they're all free to join.   I can now select any of them (one or multiple) and join them all at once.

Remaining issue

One thing I noticed a few times is that some of the data appears to be mixed up.  The only indication I've had of that was doing some spot checks on profiles. So I would grab a random profile from the database, look it up in-world, and compare the various things gathered. They all matched, with the exception of the pictures. And that doesn't happen often either, but once in a while, someone is allocated a picture that isn't from their profile.  So this is one of two things. Either people happened to change their picture shortly after being seen by the bot, or something isn't quite right when i look up pictures, and I'm grabbing the wrong one on occasion.   Because I cannot replicate the issue at will, it makes it very difficult to track down. And since it's only happening with the pictures, I'm not overly concerned.  The pictures were there more or less added to the data as an entertainment factor. As I've said before, even tho the picks are linked to the avatar, I'm not really interested in the people's profiles nearly as much as I am interested in the picks or groups listings.

Feedback

In the past week, I've received a few messages.. Either they were sent to the bot directly (guys, stop flirting already! And your pickup lines? You gotta work on those...) or they were sent to me directly.  And so far, it's been a healthy curiosity.  People are curious to know the progress, how fast the data is accumulated, what the end date might be... they have more questions after reading the blog, or they haven't read the blog and they have questions.. It is great feedback, and I will be implementing a few things that will provide that information automatically, and that part can be made public, either directly in-world, or on this blog (when I get around to making it work).  For the curious, here's a quick picture of my current dashboard:


As for the questions, I'm quite opened about it, and most questions are either answered in IMs, or I will redirect people to this blog if I've already explained it.  

 The privacy issue

Another feedback I've been getting is in regards to the bot's visits.  I have had only one person so far that seemed rather upset that the bot showed up unannounced and uninvited, and more than once, at the same location.  I've been told to respect people's privacy.  I want to address that.

Look. I can appreciate that you don't want to have a bot hanging out at your place.  And that was taken into consideration in the design and further adjustments that have been done.  When the bot shows up at a location, it typically won't stay more than 10-15 seconds, unless there is a very large group of people on site, in which case it could take up to a minute before all the data is gathered and the bot goes somewhere else.  I'll say this again, this is no different than if I, as a human being, clicked on things with my mouse, and landed at your location, looked around for a few seconds and moved on to a different location. It's not damaging anything, it's not harassing people, it's not spamming the crap out of other avi's (unlike some locations where you have the spend the first 5 minutes dismissing notifications, and group invites, and rules and regulations, and welcoming messages, and rate cards and the list goes on).  So should you feel inconvenienced by it, please feel free to express your feelings. Just know that there is very limited chance I'll do anything about it. If it bothers you that much, just block the bot from entering your parcel. 

Looking at the IMs received by the bot, if we skip all the corny pickup lines, the main request I've seen is someone telling the bot to move somewhere because it landed on the rezz spot.  Sorry, I have no control over where the rezzing happens.  It'll go wherever the pick was pointing to.  And if your landing area is in the middle of your dance floor, that's on you. I can't fix that.

And lastly, the privacy issue.  I have mentioned this before, but I think some people are overly worried about privacy issue the wrong way.  So I will explain this again. The information you put on your profile is public data.  If you don't want people to know what groups you're on, hide your groups. If you don't want people to see your face, put a puppy picture on your profile. If you don't want people to know your ASL (Age Sex Location) don't put it on your profile.  If there is private information out there, it's up to the people publishing that information on their profile to do something about at the source.  In RL, if I don't want people to know about my hobby, I don't ware a poster on my back saying I love blues music. But if I wear a t-shirt that says I like blues, and someone walks up to me and starts talking to me about blues, I can't scream that they are invading my privacy, I am the one advertising it.  SL is no different.  If you advertise it on your profile, it's public data. Period.

On a better but more technical note...

Over all, the reception has been fairly positive based on the comments I got. So thank you for your understanding, curiosity and appreciation. It is a very interesting adventure. I sometimes look at the progression as well, and it's a funny, yet expected, pattern... Groups of people share common data.  So it is not unusual to see the same picks or the same groups linked to multiple avi that are in one location at the same time. So looking at the screenshot of the dashboard, above, the numbers are a bit misleading, especially when it comes to the Picks themselves.

As noted earlier, people use Picks as an extension of their "About" section. So a lot of those picks either point to an invalid destination, or, as seen more often, people will duplicate an existing Pick (with a valid location) and change the text. The end result is multiple picks that go nowhere, or that all go to the same place.  I have no way yet to know it goes nowhere, but the "same location" could get cleaned up at Teleport time.  It wouldn't be   too difficult to compare with previously visited locations to see if it is matching one that has been processed earlier. The question is, how can I accurately determine that?

When retrieving Picks information, there is very limited data that can help identify duplicates.  Primarily, it is by location. Either Global Coordinates, or Local Coordinates with the sim name. So, how can I safely assume that one set of coordinates is the same as the other one? Coordinates are in X, Y and Z format. X and Y will provide the location, the same way you would on a graph paper. The Z refers to the elevation.  So let's say I pick a location in the Porvoo sim, at 98, 142, 113 (the bot's official residence), what does this tell us? Well, let's compare this with the RL world. A simname would be like the city name. and the X, Y would be the physical address in that city, and the Z would be which floor of the building you're looking at.  So with that, how do I determine a location is unique?  If I go by X and Y alone, I might be missing that other locations might be in a skybox.  So, the simple and easy way is compare all 3 values right? Sure, look at the values I listed above. Now, change this to 94, 152, 107.  Is this a different location? Not really.. Same house, but this is the main floor, on the front porch of the house.  So how would I determine the boundaries of a pick ?  This is the kind of challenge I'm having in the cleanup of data.   Pieces of land (Parcels) are of different sizes. So even trying to go by range (+/- 100 in each direction?) could potentially miss locations. Especially in commercial areas where you might have multiple unrelated shops side by side.   Or maybe a club has some stores attached to it. Some picks might refer to the club, some might refer to the store.. .They are right next to each other, so how do you set the boundaries?

Closing words

This post is a bit longer than I originally planned, but I think the information provided above might provide another angle to look at the challenges encountered.   The goal is always to be as little intrusive as possible, so we don't affect other people's enjoyment of their SL, while at the same time gathering as much information  in the shortest amount of time as possible.

I hope this post was valuable to you.

 


Comments

Popular posts from this blog

The first post is always awkward

A bit more on addictions...