The third submission deals with an interview on IRC between Dan, Knight, and a 'well known' ex-AOL tech. His identity is verified by our first source, but we can't publish it due to reasons that shall not be mentioned. This interview deals with more technical aspects, and as it isn't quite finished yet, will contain more material at a later date.

Questions asked by Dan can be found in bold red, while KnightHeart's questions are in bold blue, and the source's answers are in regular black.

"In what capacity did you work for AOL?"

I worked primarily in technical support; I also worked several months doing saves as well. Additionally, I was part of a special project that you might ask me about later. Suffice to say that I was off the phones in something of a supervisory capacity.

"How long were you there?"

I worked there for just over a year.

"Why did you quit?"

Well, the most immediate factor would have been my last boss. He was something of a tyrant. However, it's really a mix of circumstances that led up to it. I was tired of the day-in day-out routine of having to deal with people (both management and members) that didn't really understand what the service was, what we were there to do, and no one blessed with any idea about how to treat others.

"We've heard about AOL's Borg mentality...any thoughts?"

It's been heading in that direction for some time. There's a certain amount of that attitude that actually resulted from something that would have been quite positive. One of the problems that we had (company wide) was that there was very little consistency. Nobody did the same job in the same way, and that was a disservice to the members. They would call in time after time and get so many different answers that it was nearly impossible for them to trust anyone.

"Why is that? Was this before Sherlock?"

Having some amount of consistency in the way that people did their jobs would have helped quite a bit, and that's what Sherlock COULD HAVE done. Even to this day, I don't believe that it accomplished the consistency that was so necessary. Instead, it turned the job into more of a guessing game than it already was with Swift Response (the predecessor to Sherlock).

"What are your thoughts about the effectiveness of Sherlock? Good or bad?"

The bad side of the Borg, of course, was that it took away much of the ingenuity that had existed. It forced it out in ways that were rather disturbing. Sherlock was terrible. Even after 4 months of testing and tweaking, it still couldn't do half the job that it needed to do.

"What made those ways 'rather disturbing' to you?"

The developer said that it would learn, and as it learned, it would grow to have the knowledge of the whole company. It would be larger than the sum of its parts in other words.

"So they actually thought of Sherlock as artificial intelligence?"

The most disturbing was to see all the good techs leave. People were there that had a fair amount of knowledge. They were demoted, promoted to other positions where they would be incompetent, or were simply fired or quit. Sherlock was touted as being an AI database, yes. In a way, it was. It was AI because it depended on us (all of us, even the bad techs) to teach it.

"From what we've heard, Sherlock does nothing more than regurgitate a bunch of options based on the first five words a customer says. True?"

Yes and no. It was meant to be that simple, and the people who tested it said that they were told this very same thing. Still, everyone learned how to interpret what the member said ("member speak") and put it into Sherlock to come out with an answer that would make sense. It was supposed to be member speak only, but then ridiculous things would happen. Someone "couldn't get mouse to work on the Internet" and all of a sudden you were busy asking them questions about whether their modem was attached to the computer. Silly things like that were fairly regular when we started using it. I would bet it's better now, but not because the AI has "learned" but because the techs have learned how to interpret "member speak" and type in something that actually works.

"Well, that figures. On techs, what is your opinion of those that AOL hires as tech support?"

They are getting progressively worse. That's part of the reason that Sherlock was necessary. I loved the old Swift Response database (regardless of it's poor reliability) since once a person learned Swift, troubleshooting was much faster, and could be done from memory after a couple of months of experience. However, I was a bit more technically inclined than most. We were told (and I believe this), that AOL ran dry of technically proficient individuals to hire about 2 or 3 years ago.

"So instead of hiring competent techs, AOL is basically hiring any dummy that can type?"

The employees that they would want to hire were simply not there, and it forced them to hire Joe Blow off the street who couldn't necessarily use a mouse or type at a keyboard. It was more difficult to teach Swift Response since it was not an intuitive database and didn't allow for the fudging that Sherlock does. Also, there were fewer people who knew Swift well enough to teach it effectively, and at the time, much less time to do any training at all. Don't get me wrong, AOL brought this on themselves. They didn't plan effectively. They didn't know how to grow the company at the same rate that they were throwing millions of seeds on the dirt.

"So, in your opinion, today's AOL tech staff are nothing more than poorly trained monkeys told to let Sherlock do the thinking for them?"

It's a catch-22. They are indeed poorly trained. Their trainers (even those that know some tech) end up spending the majority of their time on customer service training. They say that Sherlock will make up for everything else, and since it can't (because it depends on knowledgeable techs to feed it), they are left in a hopeless position. In fact, the AI is smart enough to remove case resolutions that aren't used, making it an even worse situation.

Again, back to the consistency issue, Sherlock should have/could have made things more consistent, but the AI itself changes things on a weekly basis. Maybe last week you did a DNR, but this week, Sherlock just says to delete a certain file and be done with it...or vice-versa. So much for the consistency.

"Heh...ready for some AOL 5 questions? What is the worst thing about AOL 5.0 that you know of?"

Go for it. Just a small disclaimer for myself though: I wasn't around AOL when 5.0 went Gold Master. I did some beta testing with it, and also have some experience with installing it since I left...The worst? I would have to say the worst thing about 5.0 would be that it has streamlined the install process so much (a good thing normally) that it is almost unusable in any situation where you are not the "typical" AOL user. If you are only going to use AOL, and have no other apps or ISPs, or what-have-you, then it's okay, for anyone else, it's just asking for disaster.

"How have you seen on AOL 5 screwing up people's systems?"

Heh, yes!!! I was awakened early one afternoon (being a typical nightowl) to find that my roommate at the time (another victim Steve Case's employment tactics) had installed 5.0 and promptly could not ping, could not telnet, could not traceroute, could not see any other computer on the network, etc. I was charged with fixing it, and sadly, even after an uninstall of AOL, a reinstall of the TCP/IP stack, and other twists and turns, it boiled down to "format C:".

"And why do you think that is? Does AOL overwrite vital files with its own, inferior version?"

Since then I have found that many problems (by experts) can be solved with AOL, but none of the solutions are intuitive, nor do I even know if the AOL techs know of any of them. It does overwrite many files and settings. Part of this was an attempt to fix the perpetually breaking AOL adapter. The AOL adapter (with 3.0 and 4.0) came broken. Contrary to what many others told me, the AOL adapter came with settings that would only work for a certain amount of time before self-destructing in a burning pit of "connection to server could not be established...." They tried to fix it with Matterhorn (the second release of 4.0), but even that came broken much of the time. I imagine that they wanted 5.0 to take care of it more permanently. Still, 5.0 has a nasty tendency of overwriting files (I can go into the mechanics a bit if you like). It's all for ease of use, but probably hurts more than it helps.

"Figures. Who would you blame for this, the programmers or the CEOs who want to rush out new versions?"

Both. The programmers knew what they were doing. Some of the ideas were truly spectacular for making the installation easier for everyone, members and techs alike. Nonetheless, they didn't take the good ideas far enough. They didn't provide a way out for the experts. The "suits" (as I will here refer to them) also rushed the programmers. 5.0 had to be out well before Christmas in part so that the bugs could be worked out before everyone got their new computers, and there was also a lot of pressure just because Member Services was so bogged down and overworked that they were hoping 5.0 would bail them out.

"Ok, so what are some of those mechanics you mentioned?"

AOL (all 32-bit versions) install an "AOL adapter" into the network control panel. This adapter is meant to be a piece of "virtual hardware" much like a real network adapter. In particular it creates the interface between the AOL client and Internet Explorer. I presume, as well, that it also permits the binding to the wsock32 (allowing for other apps to tunnel through the connection). In order for this to work, the TCP/IP protocol must be installed. For the most part, Dial-up networking comes preinstalled on nearly all systems, and TCP/IP right along with it. When AOL installs, it comes with a set of files contained in a couple of its own directories that contain copies of all the network files usually already present on a system. In some cases, it will overwrite (without asking) other files regardless of whether it overwrites a newer or older file in its place. The problem is one that all programs potentially suffer from (but most programmers are smart enough to avoid): you might need some of those files that the AOL installer is overwriting, and unless you are a crack computer user, you'll never know what happened.

"Isn't Windows itself responsible for checking whether or not files being installed are newer or older?"

Given half a chance, Windows will do it. For example, uninstalling the AOL adapter will prompt the system to examine its files; then, you'll get prompts about any files that have changed. The basic problem is that quite often, those messages are suppressed during the AOL installation. Maybe Windows would do it, if it could. It may all be coincidence, but AOL's proprietariness seems to exceed even Microsoft's.

"So AOL is actively suppressing the capability of Windows to alert a user to the potential problems of overwritten files?"

I have no other explanation for it. I've seen installs where it did ask, but most of the time, I only see the prompts when I'm busy fixing something that has already gone wrong.

"So AOL's programmers have willfully made it so a user's system is secondary to the AOL program itself?"

In some ways they have. That's their battle against Dial-up Networking. As the argument goes, DUN involves an extra several steps (configuration and complication) that isn't necessary. AOL comes up with their own setup that circumvents the DUN. Unfortunately, it can't circumvent the networking setup of Windows in a way that would allow other (third-party) applications to work. So the solution is to pass on the DUN, but figure out how to get along with M$ networking setups. It's no easy task, and in some ways, 5.0 has gone backwards in this area.

"What do you know about the reliability of AOL's various servers?"

Email usually got where it needed to...eventually. I started there during the email server change. It was so funny to talk to 7 or 8 people everyday that couldn't get their email flags to go down. After that, things were good for a while, though it was not unusual for 5% of mail to be unavailable for a couple hours everyday. The web proxies were good, except for members who used Macs. I never dealt much with the news server when I was there. People didn't call about it most likely because they didn't use it. Some of the various keyword areas would go down on occasion. Some sporadic problems were either never solved or solved only after 2 or 3 months.

It was a rare occurrence, but occasionally bizarre things would occur that defied reasonable explanation. These problems were beyond Member Services resources to troubleshoot, and our upper-level support was unfortunately rather non-productive. That made for a lot of floor lore about various fixes that weren't in Swift (and then Sherlock). A simple one that comes to mind was the use of KW: quickmenu to circumvent the idle time, the 45 minute timer, random disconnects, etc.

"The ones AOLamers would think of as 'haxor' fixes, eh? :)"

Yeh, ways to get around the service. Moreover, they were ways of getting off the calls when they started going rather long. People perceived that things were better probably because they saw that they stayed on for 5 minutes longer when they first used Quickmenu. They remembered those instances and just thought of the other times (even when they used Quickmenu) as flukes when it didn't work.

"Sheesh. Okay, last question for me. What motivated you to speak out against AOL?"

It was a travesty to work for them. In some ways, I let them destroy my life. I thought I would go places with them--climb the ladder. When I ended struggling, there was almost nowhere to turn. By the time I quit, I felt worthless. I hardly cared what I looked like. I was a slob, and while I can't exactly blame them for that, it was the mentality of a great many people there. The personal stuff aside, we were expected on a daily basis to shove people off; it was expected to transfer them to a telemarketer. It was expected to provide them with only one resolution per call, and if they needed more support, they would have to call back. We were asked to treat people as if they could not learn (and in fact, the developer of Sherlock told us that the members never would learn--sounded like Adam Bailey in many ways). It was inhumane the way things ended up. And, I'm given to understand that the all-hands meeting (held shortly after I left) informed us about how we needed to get rid of the "me-disease." As if we were not already subjected to losing ourselves to the company, we were explicitly asked to not be individuals.

"'Be all for AOL, be nothing for yourself,' huh?"

The stock was supposed to be our reward...

For the rest of this interview, click this link.

Note: All text, graphics, HTML code, and sounds are exclusively the property of Anti-AOL.org unless otherwise indicated.
(c) Anti-AOL.org 1999, 2000 all rights reserved. Click here for more info.