Transcript of Currents 013: Rob Malda on the Slashdot Story

The following is a rough transcript which has not been revised by The Jim Rutt Show or by Rob Malda. Please check with us before using any quotations from this transcript. Thank you.

Jim: Today’s guest is Rob Malda. Back in the day, he was known by his Slashdot handle, CmdrTaco, and he tells me he’s still known by that. He was the guy, along with one of his buddies who started the famous website Slashdot. Way back yonder, it was absolutely must read when I was CTO of Thompson, now Thompson Reuters, I’d say about 1999 onward, I checked it every day when I was CEO of network solutions and we were involved with the [ICANT 00:00:33] wars. It was again, an absolute must read. Interesting how I happened to invite Rob on the show, in fact I’d call it some synchronicity. A couple of weeks ago, I was talking with my friend Peter Wang, who’s the CEO of Anaconda the Python tools company, about moderation systems of the past, and what they might have respect to lessons for the future. I mentioned then Slashdot, and mentioned it had a very interesting and unusual rating system going on. Frankly, I said, “I had no idea whether Slashdot still existed.” Peter assured me that it did.

Jim: Then last week, I saw Rob musing on Twitter about writing a history of Slashdot. While I’m supposed to be on a six month break from social media, where am I now? I guess halfway through month three, I still do one screen a week just to get a zeitgeist of the vibe on the net. I just happened to see that, I guess one of my friends happened to like it or share it or something. I said, “Shit, two signals in a row in seven days, got to be something there.” I reached out to him, said, “Hey, Rob, hey you can start talking about the history of Slashdot here.” Rob, how you doing?

Rob: I am well.

Jim: That’s a good thing in this craziest of all year. Let me start with the obvious first question, where the hell did the handle CmdrTaco come from?

Rob: CmdrTaco is a reference to a Dave Barry quote, he wrote a book, I don’t know, probably in the ’80s, maybe ’90s and it was like a joke dating guide. The Commander Taco was the name of a restaurant that you should never take a first date.

Jim: In my hometown, it was quite common to take a first date to Jack in the Box at two o’clock in the morning for the nastiest, greasiest tacos ever. I guess we didn’t read that book.

Rob: There you go.

Jim: How about that? Next we’re going to talk about Slashdot today. How did it start? What was the deal? What the hell were you guys doing when you launched it.

Rob: Yeah, we didn’t know either. Slashdot started, let’s say 1997, it grew out of, I was working at an ad agency as a part-time job building websites. I was in college at the time, let’s say 1995, 1996, nobody knew how to write HTML. I hung out with graphic designers, I was an engineer and I hung out in IRC chat rooms and Usenet message boards and was writing what would in 2020 be commonly known as a blog, but in 1996, 1995 had no name, it was just a homepage. All of those things just sort of swirled together and over the course of a couple years there, I built the Slashdot thing.

Jim: Cool. As recall, the software that you ran on, has always been open-source, was that correct?

Rob: I don’t know if it’s always been open-source but it was definitely open-sourced after a time. The first couple of versions of it were not anything shippable. They were hodgepodged Perl scripts duct taped together. But after a few years, we made a conservative effort to release the whole platform as an open-source package. Lots of websites in the early ’90s ran it, it was called Slash. There was a number of sites that were of varying degrees of popularity that used our code. Although, I don’t think that any of them really plumed the depths the way we did. It wasn’t really a useful application for most people because it was itself highly tuned for our specific use case which was a pretty large population. A lot of the weird code that made it difficult for newbs to install was basically built around us building a scalable platform, both in terms of the available hardware that we had, average folks didn’t have 15 servers to host it on. But also, we were constantly dealing with trolls, so a lot of the code in the system was the moderation system. The system was overly complex if you didn’t have our problems.

Jim: Yeah, I know that feeling. Back in 2013 I was part of a project where we put up a community system board that we thought was going to get bigger than and it did and we foolishly took a fork of the Reddit source code.

Rob: Oh, you [crosstalk 00:05:20]-

Jim: And it had the same issue, it had massive backend scaling potential, which we thought we were going to need. Well, guess what? We missed that by two orders of magnitude, oh well. It was a disproportionate pain in the ass from a DevOps perspective and from an operational perspective. But it was fun nonetheless. I actually still have that code, I know it’s no longer available. So if anyone ever needs a copy of the Reddit source code from 2013, including a bunch of cool enhancements, let me know.

Rob: I’ll keep that in mind should I ever find a desire to build up another site.

Jim: The code is not bad, but I would also say it is not good. It’s somewhere in between if you know what I mean.

Rob: I have a feeling that it was developed under not dissimilar circumstances from my own. That’s not really the ideal way to start an open-source project.

Jim: As far as I know, they never rewrote it, you can just see the… it’s kind of like the ruins of Troy, 13 layers of slag, one on top of the other. You started this about 1997, when did you start to build significant traffic?

Rob: I guess I don’t know how you define significant. We were pretty big pretty fast, but I guess that’s all relative. I was very active in the IRC community, for the Linux space, I was an open-source developer, I was writing code and contributing to various platforms. My site, even before Slashdot existed, had a certain amount of traffic. But, that would be like hundreds or thousands of users a day. Slashdot itself, when it came out, when it started was just me saying all of my stuff is here now. It had thousands of people almost immediately. But it grew to tens of thousands in a few months, and hundreds of thousands within a year or two.

Rob: We were very fortunate, I don’t have a lot of advice in terms of how you would grow such a thing except for to do the right thing and don’t be dumb. Seems like that’s a lesson that a lot of websites don’t really understand anymore.

Jim: Hey, let’s also be honest, be at the right place at the right time, right?

Rob: 1997, man.

Jim: Exactly. As I said in the opener, it was kind of the must read source for serious biz tech nerds in the day. Whenever there was some hack or some new policy issue that was arising, first place I went, Slashdot.

Rob: For a good five years there, we were, I wouldn’t say the only game in town, but we were head and shoulders above the rest. Because your other choices would be the mainstream tech publications and they very much grew out of the traditional print media, so they had a certain flavor to them. I wouldn’t call them stale, lame and boring, but I just did.

Jim: Other people might. It was interesting, but then on the other hand, eventually started to fade, I still sort of recall when I’d first noticed it fading, by this point, I’d retired from business in 2001 and had reinvented myself as a scientist. I was out at the Santa Fe Institute doing research on agent based evolutionary artificial intelligence, how about that for a mouthful, but I still had some tech investments and still advised, I was on a few boards and what have you. I remember something came up and I immediately went to Slashdot and said, “Slashdot should be the place to learn about this.” Must’ve been about 2003, 2004, and there was hardly anything there, and what was there, it was a little weak. Then I discovered Ars Technica, that then became my kind of go-to place around that time.

Rob: Yeah, ours was solid for a long time there. I would say Slashdot held on its relevance up until the era of Digg and Reddit. Once they started achieving a certain amount of traction, Slashdot lost some of its luster. But there was a lot going on and the new, I guess the more modern crowdsourced flavors became more popular. Along with that, I think Slashdot began to gently set into the sun.

Jim: Yes, that happens sometimes. Anyway, let’s get to the point that Peter Wang and I were talking about. At least back in the day, as I recall, and who knows, maybe I’m getting senile, but I recall that you had a rating system where people were selected somehow, in relatively low frequency, to rate the stories. Is that correct, did I remember that right?

Rob: Well, I might nitpick on your terminology. Slashdot was always a… The story selection process, when you say stories, I think about the story selection process, and I think you’re trying to describe the comment moderation process. We always took story submissions from our users. In the later era, the 2007 range let’s say, there was a voting system where the community could participate in the story selection process. But, within about a maybe a year and change of Slashdot’s existence, the comment moderation system was very much an active development. I think that’s probably the piece that most people think of when they think about Slashdot and its role in large scale discussion systems. There weren’t a lot of discussion systems in 1998, 1999 that would potentially have hundreds of active contributors posting within a few hours, and also allowed anonymity within the context of the platform. Our system was uniquely evolved to deal with the problems that we saw.

Jim: Okay, I think you’re right. Again, this is a little while ago and I’m getting old God damn it. It was the comment moderation that you picked people, it was not everybody got to vote, people were picked for short periods of time, like a couple of days, and you had… I don’t know, did you have a limited number of votes? I don’t even remember.

Rob: Yeah.

Jim: But I still remember being selected several times and taking it seriously, because it was high powered signal. Unlike on Facebook or an upvote on Reddit, which you think of as a cheap signal. If you got tapped with this limit period franchise, you actually took it seriously. Why don’t you explain how that system worked?

Rob: Yeah, scarcity creates value. The way that the system generally worked is that you had to be an active participant within the community, which I had a number of different metrics that I would measure, to decide if you were able to participate or not. One of the things was, “You got to read the site and you can’t read too much.” There was sort of a sweet spot. You had to read the site more than a couple of times a day, and you had to read the site less than, let’s say, I’ll pick a number, 50 times a day. If you were that, then the concern would be that you were trying to game the system and get points intentionally, and we absolutely had to deal with that problem later on.

Rob: But the system would basically dole out eligibility tokens if you happened to be, if you fell within the window of people who were good contributors during that time window and when the system… Basically, as users would post comments, we would push eligibility tokens into the system, and once enough comments had been in the system, we would pull whoever had the most tokens out of the pile and give them mod points. That would actually amount to a few hundred people a day, or maybe a few thousand people a day depending on the era, getting mod points. At which point, the specifics would vary, depending on whether you’re talking about 1998 or 2008, but you’d get, let’s say five or 10 points and you were usually given 24, 48, 72 hours to spend them. At which point, you would read the comments, and you would find the ones you liked and you could choose to send them up a point or down a point. At the core, that’s the whole system.

Rob: We had scarcity which created urgency and over time, we were able to identify users who were bad actors and users who were good actors within the system, and sort of adjust some levers to make sure that the right people were getting the points at the right time, and the bad actors got weeded out of the system. We were able to get to a place where we were able to generate a community that could spend thousands of points within the system a day, which was usually enough. I tried to maintain a ratio of comment posts to moderator activity, and that kind of was the sweet spot. Whereas, today when you see a tweet or a Reddit post with your likes or whatever, that’s essentially an infinite resource, you can have a tweet with 10 million likes on it but the Slashdot system was designed to not actually have that, because I always felt that the scarcity is what made the system kind of reliable.

Jim: Yeah, exactly. At the time, when I was most active on Slashdot, I was the CEO of a billion dollar company. My time was precious. But when I got those little taps, I took it seriously, and I actually said, “Okay, here’s some significant signal in the community that I respect.” So I did it and I paid a lot of attention to it. I think that was a damn clever design. The other thing that’s nice about it, is it defeats some of the game theoretical hacks on open rating systems. Famously Reddit is where brigading got started. I can still remember the shit Reddit says versus the anti-shit Reddit says, brigade wars when hundreds of people would attack each other posts and up rate them or down rate them, et cetera. Your system seemed pretty damn resilient against brigading as an example.

Rob: It’s interesting is that, I said this earlier, but in the later ’90s we developed a system to rate stories, and to sort of leverage our population in thee story selection process. But I wasn’t comfortable simply flipping the switch and using a model like a Reddit or a Digg where the community votes exclusively to determine the content of the homepage. When that happened, we saw a lot of that brigading style attacks on this new system, which didn’t have the same scarcity of moderation. But, I had designed that system to avoid it rather than revealing for example, specific integer values for the posts. I color coded them so the color codes underneath the hood could represent any number, it could be a million likes or it could be seven, if it was the right seven. If I moderated it up, that was worth more than you, because sorry Jim, you’re a nice guy but I know what I like and I know what’s going on Slashdot.

Rob: But the system, we absolutely had that brigading type behavior in that system and the development of that story rating system, very much was influenced by the successes that we had had with the moderation system. I was trying to find a hybrid between the more public Reddit type systems, which are easily gamable, or relatively easily gamable. What I found actually, is that some of the worst offenders for the brigading style tactics were actually the publishers themselves. They would routinely get into our system, we called it the fire hose, they would get into the fire hose, and they would moderate themselves up and they would send their rivals down.

Rob: The worst would be an embargoed news tech release, where you’ve got five, but you got your Ars Technicas and you got your PC Magazines and they’ve all got the same embargoed new story coming out at 12:01, and they would all log in and moderate their own story up and their rival publications down. That would just be like a dozen people, but it was very much… I think that was kind of the era where I realized, “Oh great, the publications are just as crappy as the randos.” It took me a little while to realize that the governments would be doing the exact same thing and that’s where we see the Twitters in the mid-2015s or the Facebooks, where now you government actors what might’ve been being done by AnandTech or something in 2007.

Jim: Yeah, absolutely. Again, this is what you should expect, game theory. You give agents an ecosystem and they will attempt to fucking manipulate it. I’ll confess to, in the early days when I was, I wouldn’t call it the early days of Reddit, but the early days when I was on Reddit, I had a dozen sock puppets. Quite handy to kick your posts up a little bit, particularly in not so trafficked subreddits, right?

Rob: Sure.

Jim: You give yourself a dozen upvotes over six or seven hours and it definitely helped. Of course people are going to do that.

Rob: Yeah, the only way to get your accounts to stop working on Slashdot.

Jim: Yep, exactly.

Rob: Plus the thing is that we spent years developing defensive techniques against those sorts of behaviors. I actually felt a little bad for the Reddits and the Diggs because when they started gaining traction and popularity, those sorts of behaviors just became immediately obvious. The users of those systems, the comment posters had developed these techniques on Slashdot, and we had spent years hardening ourselves against them. When the next generation, the Web 2.0 generation came along, I felt bad because the troll population was coming to the battle with machine guns and they were defending themselves with 2x4s and sticks.

Jim: And they still are. Frankly, you can still manipulate both Reddit and Twitter fairly easily.

Rob: Although, I would actually argue that now, it seems like the people who are working on the bots are as much… they’re working with psychology as much as anything else. They know what stuff is going to bite on people and they’re really careful about that. I think that 15 years ago, it was less about the psychology and more about just trying to make your point heard, and now it’s a lot more savvy, a lot more malicious I suppose.

Jim: Yeah, it’s another game that’s opened up. The old game of up voting still exists, brigading still exists. I know a guy who has an army of trolls for instance, and he’s real active right now manipulating things in the presidential elections. There’s robot posting for sure, but good old brigading still exists, and it’s as always, an arms race, the platforms are trying to detect it. But hey, you run it through Tors, they don’t know where the hell you’re coming from but then they say, “All right now, people’s IP addresses are evolving in a nonstandard fashion, that ain’t right.” Then you start using Ghost where you can come out where you want.

Rob: You’re giving me flashbacks because what you are basically describing was my day-to-day job from like 2000 till 2006, 2007, “Oh great, they figured out a new way to obfuscate their identity.” Slashdot at its core, we really worked hard during my tenure to maintain the ability to post anonymity or to post anonymously. That’s not something I think that exists anymore, and whether or not that’s a value is a different discussion. But, we valued it at the time, and that made it very difficult for us to have a lot of defenses against folks. Because like you said, Tor or even before Tor, unsecured relay servers, that was our first problem, unsecured proxy servers. Then that would evolve into Tor and then we had people, they would bot the creation of user accounts so that they could get around… We would throttle anonymous users maybe a little bit more rigidly than regular users because I couldn’t tell if you were one user or 50 without enough data. I don’t know, it was difficult time and it gave me a good sense of what the folks that are at the major platforms are dealing with today. It’s a bummer because they got overwhelmed and I think that now that everyone’s aware that that’s a problem, it’s a little bit too late, the cat’s kind of out of the bag for those folks.

Jim: Yeah, sort of. But nothing says they couldn’t change their moderations policy. Go ahead.

Rob: Well, [crosstalk 00:23:08] moderation policies but it’s about user expectation and I think that, but users of these systems… Here’s a dumb example, the fact that the like count on Twitter is displayed as a numeric integer creates a whole number of perverse incentives for bad actors and for good actors, but they can’t take that number away in 2020, people would freak. That number is probably damaging to them. If they simply replaced that with percentiles and said, “This comment’s in the 99th percentile for today,” a lot of edge case behavior would go away. But it also, at the lower levels, would take away incentive because moving from the zero to the first percentile is actually a pretty big burden for most online content.

Jim: Yep. Though Jack has mused about getting rid of the counts, he’s talked about it and at one point, said he was going to. He didn’t actually but-

Rob: Yeah, wonder what-

Jim: … they’re at least thinking about it. Right?

Rob: Yeah, [crosstalk 00:24:18]-

Jim: Yeah, a couple-

Rob: … all the time. I think that they’re all doing their best in a questionable situation. But these decisions are hard and they have influence at the street level where you just have a random three people tweeting back and forth at each other but also at the state actor level. These are hard, hard problems and if you haven’t been neck deep in it, you don’t understand the complexities of them. I wish them all the best, I wish they would’ve hired me for like one day of consulting in like 2008 so I can sit down with them and say, “Look, this is what’s going to happen to you, you need to harden against this now.”

Jim: And again, they should’ve just known that any ecosystem that has targets of value that can be exploited, will be exploited. You discovered many of them, though probably not all of them.

Rob: Oh, definitely. We were discovering new ones right until the day I left.

Jim: Yep. Is there anybody out there that’s using the high value scarce moderation technique that you guys had?

Rob: Yeah, no I haven’t seen… Well, nobody’s using our specifics. I imagine there are systems out there that use them but they’re probably the notions that were playing with probably have more in common with video games than a moderation system. Things like running out of ammo and stuff in a game has a lot to do with, “Do we add more ammo and more weapons to make this map more fun or does that make this map be imbalanced towards a certain play style?” I think that the gaming community probably deals with these sorts of problems more than the Reddit communities. Because I think that the larger scale systems, they’re so focused on maximizing time on site that the notion that you would throttle an interactive behavior is a difficult one, because what are you doing except designing metrics and measuring success with your various metrics? What tells you that a user is enjoying your system more than them clicking a like button? That’s a big deal.

Rob: They have a different incentive. I don’t know, it’s a hard problem. I’ve never seen anybody deploy things to that extreme scale that we did. There are plenty of systems that restrict moderator access, but they generally do so to tiers or classes of users and not to… The Slashdot system was very much designed to give everybody a shot, like, “I’m going to test you out. Welcome to the system, you’ve been around for a couple of months, I’m going to test you out a couple of time and if you prove yourself worthy, then you’re going to get to participate in the system every couple of weeks. If you don’t, then you’re not going to have to play anymore.”

Jim: Yeah, it was the great idea, it really was. That’s actually the conversation that Peter and I were having said, “This was an actually interesting, different and at the time, seemingly effective way to do moderation.” We at least, have never seen it elsewhere.

Rob: I think that the systems today, they suffer with a number of interesting hamstrings on this regard as well. One of Slashdot’s key advantages in the early days is that the community was really fricking smart and the topic of a site was relatively narrow. It was all generally tech or sci-fi, it was news for nerds, it was that vein. Especially for the first few years, we weren’t talking about the general things that you would see like on a national debate stage, we were talking about a niche. If you’re an expert in operating systems, then your opinion on hardware and on video cards might be reasonably accurate. But just because you’re an expert in encryption, doesn’t make you an expert in immigration policy.

Rob: I think the modern systems suffer from, you’re expected to be all things, or people tend to portray themselves as being expert in all things when they’re online. If I’m logged in on Reddit or Twitter, whether or not I’m participating in the PC sales subreddit or The_Donald, I suppose The_Donald’s not around no more, but regardless, the system regards me as being generally equal in each of these areas, although I’m maybe more of an expert in one subject than the other. I think that that’s where we really end up in so much trouble with the Twitter these days is that you have a person who’s an expert, they have a domain of expertise, and they gain followers because they have a domain expertise, and then that allows their ideas outside of their expertise to be weighed with an undeserved value because of their expertise in another domain. Therefore, because they were successful in attracting followers from their domain expertise area, they’re opinions on areas where they’re simply not qualified, gain likes, they gain unearned popularity.

Rob: I think that Slashdot had a lot of advantage because at least in the earlier days, the community was relatively focused on subject matter. We had a good advantage there. As we broadened out and added things like politics and stuff to the site, I think that that became more of a problem. But the modern systems, they don’t have anything like that. On Twitter, you can follow me because you’re interested in seeing me make pens, you can follow me because you’re interested in my podcast Geeks in Space or you can follow me because you’re interested in my opinions on social media and modern day use of moderation systems. But my qualifications on any of these things has nothing to do with the fact, my qualification for if I happen to tweet that the Detroit Tigers won today, and obviously these are trivial examples. But people use them for legitimately important matters of discussion, and the systems don’t seem to differentiate between the two.

Jim: Yep. Then you talk about cross leverage of expertise, and of course, on Twitter, the even bigger problem is no expertise. So-called celebrities, they got nine million followers and they mouth off on all kinds of topics and they have no expertise in anything. They’re just famous for being well known, the definition of being a celebrity.

Rob: That’s the thing is that the Kanye West problem I suppose, he may very well be an expert in something that he tweets about, but he’s famous for something else and I have no way to vet his credibility on a subject that maybe he’s never really actively participated before. That’s sort of universally true. You shouldn’t trust my views on any political matter, I have no particular expertise in politics, my background is generally technology and social media. You shouldn’t trust what I think about, I don’t know, immigration policy or international relations but I might tweet about it. But if you’re voting that up, it’s not because I’m qualified.

Jim: Yep, indeed. That’s an interesting design issue when we’re thinking about how to do something like social media correct. One of the things you mentioned, which I did not know at the time, was that you had differential vote weights on Slashdot. We were considering in our 2013 system, in fact, it was kind of stupid and naïve first attempt, which we never actually implemented but we specked out making the up and downvotes, you’ll love this, weighted by the log five of your Reddit style karma. How about that for an obscure algorithm?

Rob: It’s not looney tunes, we did lots of variations on that. In the initial days of Slashdot, the scores and comments were rated from negative one to five, and the database entry was literally just an integer, and it was literally just sum it all up, add and subtract when people participated. Over time, that became not nearly good enough because you have all sorts of weird case. For example, there’s a difference between a comment that’s a score five, and then somebody tries to moderate it up to six, so it flutters, five, six, five, six, five, six. What does that mean to a user and to the moderators? I didn’t like that because I was trying to balance the number of points being spent in the system on any given day. But that’s a signal there and I have to capitalize on that.

Rob: When we designed the achievement system for Slashdot, we sort of modeled it after the way that World of Warcraft did theirs but as a joke, we made each of the tiers of your rating, of your achievement, of your score in the achievement system, each was exponential, it wasn’t log five, it was exponential. To get to rank eight or whatever in an achievement category, it had to be two to the eighth points, events that measured that. There were a number of achievements in the system that you’d suddenly find yourself, “Well, you got to post 16 million comments in order to get to the next tier because it’s exponential.”

Jim: They had a log two, that’s basically log two and log five’s even worse, right?

Rob: Right. We did essentially, a similar thing because I didn’t want people to treat that as a gamable thing. At some point, the difference between, if you’re talking about a free action, like a like or something on Reddit, if a user has done that 100,000 times over the course of a year, whether or not they do it 100,000 times or 200,000 times doesn’t really matter all that much. The difference between the orders of magnitude are really all that matters at that point.

Jim: That’s exactly why we like the log rhythmic scale, because it’s essentially the inverse of exponential. Exactly that reason. The difference between 100,000 and 200,000 posts is much different than between 20 posts and 100,000 posts.

Rob: Right, the actually comment moderation system on Slashdot didn’t exactly use that, but we definitely, by the end… The system underneath is actually pretty complicated when I left, and I can’t imagine that it’s changed much in the last eight, nine years. But there’s a lot of weird math going on underneath there. To deal with weird edge case, for example, what happens if I find a bot in the system and I discover that that bot has been doing something crappy in my system for two weeks and I didn’t notice, well, the system that we had actually had the ability to extract that user from the system, I could erase their scores. That might mean a lot of weird stuff.

Rob: Like the example I was talking about earlier, you got to score five comment that gets moderated up to six, but Slashdot doesn’t have score six, so it’s just four, five, four, five, four, five, six, five, four, five, bouncing around in that range. Well what happens when you pull somebody out, pull somebody’s influence out of that like a week later? Well that comment isn’t getting active moderation so you actually have to replay the moderation history that occurred to that individual comment and see that this guy, who you might’ve weighted him as a 100% contributing user a week ago, but now you realize he’s a 0%… You got to make sure that you keep the score fair to the user who posted the comment. It’s a hard problem and I don’t think people really think through the complexities of this stuff unless they are neck deep in it for a few years.

Jim: Yeah, and of course, it’d be even harder to implement something like that on the scale of a Twitter or a Facebook where instead of hundreds of thousands or a few millions of users, you’re talking about billions, right?

Rob: For sure. And they have the further problem where, like I was saying earlier, your post in one domain of expertise, you can get retweeted tomorrow by somebody whose followers all exist within a completely different domain of expertise. A lot of the historical value gets thrown out the window. It’s a complicated problem, and I would absolutely… I would love full access to those guys’ log files, I’d love to parch that data and see what’s going on.

Jim: Have you ever reached out to them to offer your expertise, and/or get access to their data for research purposes? They do occasionally give it out.

Rob: I have academic friends who do that sort of thing. But nobody at Twitter or Facebook knows my name so it doesn’t matter.

Jim: I had a guy, Phillip Howard on from Oxford who leads the social media analysis group over at Oxford and they do stuff like that. I had on, actually just last week, Renée DiRest, who does something very similar at Stanford. If you’d like to be introduced to those folks, I’d be happy to do it.

Rob: Yeah, you might want to talk to Cliff Lampe at University of Michigan, he does a lot of that stuff too.

Jim: Sounds good. Let’s finish the story. What happened to Slashdot? Why did Hacker News kick your ass?

Rob: Hacker News never kicked our ass, Hacker News is a different deal entirely. Well, I think Slashdot became victim of its own success. When the site began, it was a very dedicated population of contributors, who were generally expert in the areas of the nerd culture that we were doing, that we were covering. Over the years, other choices came along and when a population starts off with a core of experts, the inevitable thing that happens is that a secondary population of hangers on moves into the system as well. I think that over the course of a few years, a lot of the expert folks got pretty annoyed by the newbs and other possibilities came out too.

Rob: What happens, like Reddit and Twitter, and I guess before those, to a certain extent Digg, they created a different kind of incentive for users because the story selection process that those platforms used was driven entirely, or mostly entirely by the users themselves, which meant that those systems can be faster. A tweet that I make right now, goes out to all of my followers within a second and they’re limited only by how fast they can refresh or how often they refresh Twitter on any given day. Slashdot had itself limited to, let’s say a dozen, a dozen articles a day, a dozen stories a day, so if you want more than 12 stories a day, or if you are more interested and you want to deep dive into a specific area, well, that’s not going to work too well for you on Slashdot.

Rob: The Reddits and the Twitters allowed the power user to get more of what they wanted, and it allowed them to get it faster. Once the power users started figuring that out, well the power users started losing interest in Slashdot because at that point, they have to jump through the editorial hoop created by the Slashdot editorial process itself. That’s legit and fair and I would do the same in their shoes most likely. But I think that you play that forward for a few years and you lose five, 10, 15, 20 percent of your original audience for that reason, and then what you have left is a bit of a hollow shell compared to what you had before. It wasn’t so much a collapse, as much as it was a slow sad death by attrition.

Rob: I don’t think that Hacker News kicked our butt, I would actually say that Digg knocked us off our pedestal. Reddit did it mostly right. Twitter’s a different beast. Twitter, Reddit and Facebook are I guess the dominant platforms now. They each succeed and fail along a very specific vector, but they each have very clear advantages over Slashdot, even without talking about the scale advantages that you have. Because the other thing to keep in mind is that, with Slashdot, I very, very consciously worked my butt off to keep it within a very specific sphere of domain expertise. I didn’t want Slashdot covering stuff that was outside the news for nerds umbrella. If you want something that’s not under the news for nerds umbrella, you got to go somewhere else.

Rob: I think that was Slashdot’s advantage but it had the negative side effect that anybody who’s going to be reading Slashdot in 2008 was probably reading Slashdot in 1998. The thing about people is that we can’t make more people that were reading Slashdot in 1998, that’s a fixed pool that can only shrink. The only way that you can grow your population is to broaden your domain of expertise or lower your standards, and I wasn’t really comfortable with either of those options. I don’t know what’s happening over there today but I imagine probably more of the same.

Jim: Interesting. Well, Rob, thank you very much for a walk through the history of one of the more important websites of that era. I was particularly enjoyed jumping into more of the details on how your comment moderation system worked, and I hope some listeners out there who work at some of these platform companies or who have influence on them, might say, “Hey, maybe they ought to talk to this guy Rob. He’s been through this shit before.” May not be directly applicable but some of this thinking I suspect would help these platform companies do a better job on what they’re trying to do today.

Rob: I don’t mean to brag but I pretty much think I’m the world’s greatest and that they should all be paying me as a highly paid consultant on all of these matters, but I might have some biases.

Jim: It’s just possible, just like I think my granddaughter’s got to be the most beautiful baby in the world, right?

Rob: I have no reason to believe otherwise.

Jim: All righty. If you want to catch up on what Rob’s thinking or at least babbling about, you can catch him up on Geeks in Space podcast, available on your favorite podcast app I’m imagining.

Rob: Probably. I honestly don’t know, I’ve never heard it.

Jim: Yeah, frankly I won’t tell you how many times I listen to my podcast but it ain’t many. Anyway, thanks, Rob, and this was fun.

Production services and audio editing by Jared Janes Consulting, Music by Tom Muller at