I agree. One of the better ideas to come along in the last while. Now if some people would just get over the ZMOG! ADMIN-BOT! part and move on with their lives and realize that wikipedia needs this type of automation.

-Royalguard11

On 1/7/07, Guy Chapman aka JzG guy.chapman@spamcop.net wrote:

On Sun, 7 Jan 2007 18:18:50 -0500, "xaosflux" xaosflux@gmail.com wrote:

...

FYI: A proposal to grant sysop rights to an automated account is currently being discussed at [[Wikipedia:Requests for adminship/ProtectionBot]].

And a very fine idea it is, too.

Guy (JzG)

http://www.chapmancentral.co.uk http://en.wikipedia.org/wiki/User:JzG

---

WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l

James Hare wrote:

Automation is good, because people are lazy.

(Don't believe me? Count the number of project-space shortcuts there are.)

We should just *try* this -- maybe then people will get over this phobia.

I still think the code needs to be widely available.

-Jeff

On 1/7/07, James Hare messedrocker@gmail.com wrote:

I can imagine how that can be benefit; it'd allow for public auditing of the code.

Problem is, if you let everyone see it, you let EVERYONE see it -- even the people who will take a look at it and figure out how to compromise the bot.

There's discussion of this issue on the talkpage of the RfA.

Newyorkbrad

What's the point of showing everyone the code? Are we afraid the owner is going to abuse the bot?

Mgm

On 1/8/07, Steve Bennett stevagewp@gmail.com wrote:

On 1/8/07, James Hare messedrocker@gmail.com wrote:

...

I can imagine how that can be benefit; it'd allow for public auditing of

the

...

code.

Problem is, if you let everyone see it, you let EVERYONE see it -- even

the

...

people who will take a look at it and figure out how to compromise the

bot.

Yup. Historical consensus is that the advantages outweigh the disadvantages. See [[Security through obscurity]].

Steve

---

WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l

MacGyverMagic/Mgm wrote:

What's the point of showing everyone the code? Are we afraid the owner is going to abuse the bot?

What's the point of hiding the code? Why does it have to be kept a secret?

And yes, I know there's discussion at the RfA talk page, but "vandals might use the code" is a poor rationale for not keeping it open. There's no good reason to keep it secret, but the headcount is going in the support column regardless, and that's a shame that it doesn't look like the arguments regarding openness of the code (suddenly the "we shouldn't accept things that are less free" folks are quiet) are going to gain any traction.

-Jeff

On 1/8/07, Ilmari Karonen nospam@vyznev.net wrote:

Jeff Raymond wrote:

...

MacGyverMagic/Mgm wrote:

...

What's the point of showing everyone the code? Are we afraid the owner is going to abuse the bot?

What's the point of hiding the code? Why does it have to be kept a secret?

And yes, I know there's discussion at the RfA talk page, but "vandals might use the code" is a poor rationale for not keeping it open. There's no good reason to keep it secret, but the headcount is going in the support column regardless, and that's a shame that it doesn't look like the arguments regarding openness of the code (suddenly the "we shouldn't accept things that are less free" folks are quiet) are going to gain any traction.

For things that go into the MediaWiki codebase, yes, they do need to be not only made public but in fact licensed under the GPL. For something that Robert or anyone else is running on their own computer, not really. Of course it would be _nice_, from an ideological perspective, if all code was public and free, but in practice I see no grounds for demanding this.

In fact, I personally feel there's way too much bureaucracy going on here, and perhaps even in the "ordinary" bot approval process as well. The way I see it, it's really no-one else's business how people choose to make their edits and other actions, whether they do it manually in a browser, assisted by user scripts, with a fully-automated bot or by telnetting to port 80 on en.wikipedia.org, and whether they do it from one account or several, as long as they admit which accounts are theirs.

If they disrupt Wikipedia, they should and will be blocked and stripped of their other privileges just as readily whether they do so with a bot or in any other way. And if a trusted user should ask that an alternate account of theirs be provided with a subset of the technical abilities they have already been trusted with, why, that should be a mere routine technicality that any bureaucrat or steward should be able to satisfy upon a simple request.

It seems our approval procedures have gotten to the point where we're just creating bureaucracy for the sake of bureaucracy, and even demanding that things be discussed and !voted upon twice just so that people whose participation on Wikipedia is centered around a single project page should not be, god forbid, made to follow a link to another page in order to participate in a centralized discussion there.

(Yes, I'm aware I'm digressing a bit from your specific point. Sorry for the rant, I just felt the need to get it out of my system after reading some of the comments on that Rf-not-really-A page.)

-- Ilmari Karonen

All I can say to that is here here! I completly agree with everything said in that email. I trust to owner, so there is no reason for me not to trust them on this.

MacGyverMagic/Mgm wrote:

What's the point of demanding openness?

Because secrecy is detrimental to the project? Because the code may be useful as a basis for other processes?

The script has had a test run, it works and it fulfills a need and on top of it all it's run by a trusted admin. It would be unanimously supported if it was a regular admin request.

Would it? I couldn't possibly say that. Furthermore, the "need" is only partially demonstrated, in my mind. The vandalism on the FA I had requested for the front page might not be protected by this, and this bot may not solve the problem, but merely adjust it.

Seems overly bureaucratic to demand the code to be open source.

Oh, no, not "bureaucratic!" Come on, seriously.

One or more of the antivandal tools are also closed code to prevent vandals from abusing it and there hasn't been anyone questioning that yet.

It's not being run automatically as an administrative account. Apples and oranges.

-Jeff

On 08/01/07, bbatsell wikipedia@theskeptik.com wrote:

These steps are so tedious and error-prone that we have screwed up numerous times already, and that has been exploited. I'm not sure how many more times we can allow pictures of cleft penii or vaginal sores ON OUR FRONT PAGE without losing tremendous amounts of credibility.

With regards to yesterday's ranting, this is a perfectly appropriate example of what automation of administrative tasks should be used for... but I would see no reason the code can't be released. There's no complex checks involved a vandal can game if they know the details, simply generating a list of all transclusions from a single page and then setting protection for each. Unless the bot does something I'm not understanding, releasing the code gives no loss of security and allows the community to feel a lot more comfortable permitting it.

As an aside, every minute a vandalised image is covering the main page, we serve it up to around two thousand people (give or take 50% based on time of day, etc). Something to bear in mind.

MacGyverMagic/Mgm wrote:

That's why I don't understand some people are uncomfortable with it. It does something simple and it's not dangerous what exactly do they feel is so scary that the code needs to be released?

For me, it's exactly this kind of dismissive attitude about it. "Trust us, it does what it says. Why do you need to know?"

-Jeff

On 08/01/07, MacGyverMagic/Mgm macgyvermagic@gmail.com wrote:

On 1/8/07, Andrew Gray shimgray@gmail.com wrote:

...

There's no complex checks involved a vandal can game if they know the details, simply generating a list of all transclusions from a single page and then setting protection for each. Unless the bot does something I'm not understanding, releasing the code gives no loss of security and allows the community to feel a lot more comfortable permitting it.

That's why I don't understand some people are uncomfortable with it. It does something simple and it's not dangerous what exactly do they feel is so scary that the code needs to be released?

Because we don't *know* that. We know the programmer intends it to do something simple and routine. We don't know if he has a secret plan (he doesn't, I am sure, but someone who doesn't know of him might not accept this), or if he's made some silly mistake whereby it'll actually miss every transcluded template containing a "7", or if it'll stop checking at the sixth level of transclusion because he thinks there won't be any more, or whatever.

It's not a matter of trust; it's a matter of confidence. Because a bot doesn't have that simple and obvious "common sense oversight" option, we want to know *exactly* what it will do, not what the designer intends it to do. And the only way to do that is to see what the code is.

Fundamentally, I cannot see any reason not to release it. We're not talking a vandalhunter here; there are no complex algorithms that can be gamed, no part of this bot that will work better if its targets don't know how it works, no plausible scenario where secrecy will prove advantageous to it. It has a clearly defined, explicitly defined, method; releasing the code to show that method shouldn't, to my understanding, be a problem to its operation. So why not? It'll make people feel better about it, reduce fear and uncertainty, and let us all get on with creative work - which is what we're here for.

geni wrote:

On 1/8/07, Andrew Gray shimgray@gmail.com wrote:

...

We know the programmer intends it to do something simple and routine. We don't know if he has a secret plan (he doesn't, I am sure, but someone who doesn't know of him might not accept this),

The problem with that line of paranoia is that it doesn't trust someone enough to run code you can't see but at the same time trust them to give you the real code.

Of course, if the code that's been released doesn't match the code that's running, we can obviously note the problem. Not so if the code is held back, or only given to "trusted users," as subjective a criterion as you can get.

-Jeff

Security through obscurity.

I'm sure if you asked the guy he'll give you the code, but keeping the code concealed may be a good thing because it'll stave off the bot being compromised for some time.

On 1/8/07, geni geniice@gmail.com wrote:

On 1/8/07, Jeff Raymond jeff.raymond@internationalhouseofbacon.com wrote:

...

geni wrote:

...

On 1/8/07, Andrew Gray shimgray@gmail.com wrote:

...

We know the programmer intends it to do something simple and routine. We don't know if he has a secret plan (he doesn't, I am sure, but someone who doesn't know of him might not accept this),

The problem with that line of paranoia is that it doesn't trust someone enough to run code you can't see but at the same time trust them to give you the real code.

Of course, if the code that's been released doesn't match the code

that's

...

running, we can obviously note the problem.

If the bot starts doing things out of spec I don't think we will need the code to know that. No the paranoid arugment fails.

-- geni

---

WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l

On 1/8/07, geni geniice@gmail.com wrote:

If the bot starts doing things out of spec I don't think we will need the code to know that. No the paranoid arugment fails.

My thought exactly -- the bot is intended for one very narrow task, and if it does anything malicious or outside of that, it seems easy enough to spot. I can respect the opinions of those wanting open source code, it seems like an important discussion to have, but I'm not sure that freely open source is needed; "half-open" might be preferred, if anything. There's a longshot difference between giving the code to 20 or even 200 people, and giving it to the entire internet.

But all else aside, RfAs like this are tricky, because it mixes all sorts of concerns and issues. In addition to the usual pressures of RfA, we have a large number of people trying to write out the beginnings of an adminbotting policy on the fly; such a policy doesn't make too much sense, before we have any adminbots, but getting an adminbot through will be difficult until there's some consensus as to the exact requirements, guidelines, and such involved.

Security concerns, trust concerns, accountability concerns, those who don't want adminbots at all, those who don't want this particular adminbot, all sorts of opinions. Very complex discussion.

Tally was 146/28/9, last I checked, but it's got a lot longer to go before the discussion closes. The more community input, the better, I think.

-Luna

On 1/8/07, Andrew Gray shimgray@gmail.com wrote:

...

Fundamentally, I cannot see any reason not to release it. We're not talking a vandalhunter here; there are no complex algorithms that can be gamed, no part of this bot that will work better if its targets don't know how it works, no plausible scenario where secrecy will prove advantageous to it. It has a clearly defined, explicitly defined, method; releasing the code to show that method shouldn't, to my understanding, be a problem to its operation. So why not? It'll make people feel better about it, reduce fear and uncertainty, and let us all get on with creative work - which is what we're here for.

Well, en.wp isn't the only Wikipedia. If someone had the code, they could probably fire it up on any other Wikipedia and continue to do this until they'd hit all of them (and presumably vandalized them all). I imagine this would be far more disastrous than not releasing the code. And of course the vandals /could/ code this themselves, but it's so much easier when someone hands you the code on a GFDL platter.

Or is this WP:BEANS?

--keitei