Monday, November 13, 2017

Salesforce.com's Dreamforce '17 Wrap-up

This is just a quick post to share some of my observations and experiences from this year's Dreamforce with you.

Salesforce Tower at night

Weather
In all, the weather worked out well for this year's Dreamforce. The last time when Dreamforce took place at this time of year, it rained all week, so rain only on Wednesday night was an improvement from that -- though one not made by anyone.

New Announcements
Whether it was the timing of the event, or just that Salesforce has changed the way that they deal with Dreamforce, there wasn't a host of new features announced during the event. From a features standpoint, I think everything that they talked about is already available in the current Winter'18 release. They did talk about a new partnership/closer integration with Google, but they did something similar 10 years ago, so I'm a bit skeptical about what the real impact of this agreement will be.

Crowds
The biggest challenge with Dreamforce -- and has been the last several years -- is the overwhelmingly large crowds. Lines, lines, everywhere lines. On Monday, I attempted to go to a roadmap session, only to find a line of 15 people waiting outside a full ballroom at the Marriott. As is usual for Dreamforce, for whatever reason, the lines and crowds for the breakout sessions are much longer on the first day of the conference, then they lighten up after that. Not that it wasn't still crowded later in the week. On Wednesday, I tried going to a Google advertising session in the Palace hotel, and the line to get in ran more than the length of the hall on the main floor. I left for a bit, convinced that the session would be too full, then came back after the session began to discover that I was able to get in. It was crowded, but still a fair amount of seats in the room.

Einstein AI
This is the second year that Salesforce has been promoting their Einstein AI functionality. Einstein is more visible in Winter'18, but if you're using Salesforce Classic like we are, that functionality isn't available. This year, they've rebranded the Wave Analytics package as Einstien Analytics. They were also showing off some ability to configure Einstein's analysis parameters and build some advanced Einstein capability. For this reason, I sat through several data science and Einstein sessions, only to get to the part in one where they told you, "you need to contact your Account Exec" to enable this feature. In other words, "there is a cost to enable this, contact your AE to find out how much". Considering that Einstein is a carrot to enable the Lightning interface, I found it kind of funny to dangle the carrot, then take it away.

Put another way -- like Salesforce.com Portal -> Community or Wave Analytics -> Einstein Analytics, I'd have to say, "those are some nice features, but I don't think that management would agree to a doubling of our Salesforce costs just for that."

Creative Visualization, Design Thinking, and Drawing to Win
Perhaps the best sessions that I've sat through in a long time and probably the biggest surprises for me were sessions that I went to on Creative Visualization and visual story-telling,  Design Thinking and Drawing. Essentially, these sessions revolved around using visual imagery to present ideas and conceptual frameworks. They were an altogether unexpected bright light for me in an otherwise uninspiring Dreamforce week. With exercises like "draw the toast", these sessions were really engaging. What I found even more surprising was that many of these people came from a team at Salesforce that does exactly what they were presenting -- meets with customers and uses these visualization and drawing techniques to explain, define, and then craft solution plans for customers.

What's more, during the keynote interview with the CEO of IBM, Ginni Rometty, she mentioned that IBM has added a significant number of design teams (I don't remember the exact number). But, as you can see from this design thinking page on their site, design thinking is thematically important to IBM too. In that way, it feels like this is probably representative of a broader trend in business -- or at least in the software industry.

Dreamfest
Of course, what would Dreamforce be without the band / concert experience. As I mentioned in my post about last year's event at the Cow Palace, I couldn't imagine a band that would make me want to go to another Dreamfest event. This year, the band was Lenny Kravitz and Alicia Keys, and the event was held at AT&T park again. Last time they had the event there, the band was Greenday, it poured rain, and most people hung out in the stands trying to avoid the rain. This year, since the event was within walking distance again, I walked to the event just to check it out. Overall, the park experience was similar to the previous time at the venue -- ball-park food, VIPs directed to luxury box level, crazy crowds.

As I've noted previously, possibly the funniest thing about all of this is how surprisingly rudely selfish the Dreamfest concert crowd is. For example, there were huge lines backed up into the alcoves trying to follow designated paths down to the field. At the back end of this, the line became a mass, choking the walkway around the food stands. And yet, there was no give in the mob of people there, no flex to allow people to pass through the walkway. I've been to many other general admission events where the audience was far more polite to one another. It's something in the demographic of this audience. And, at the same time, you have to know that most of these people have been to concerts and events before, but they act like a ravenous mob being thrown a small, insufficient allocation of food -- a couple of steps away from a riot or a stampede.

Needless to say, between the food not being great, the crowd, and the music not really appealing to me, I left after three songs from the Lenny Kravitz set. Perhaps the best part of the night was walking back to the hotel, knowing that I didn't need to stand and wait with that crowd for a shuttle bus to arrive some time later than most would expect.

Keynotes
On the first day, I was on my way from one breakout session to another when I noticed a large line of people on Mission. They were queued from the corner on 3rd back nearly to the Metreon. Isoon realized that this was the line for the opening keynote that was still over an hour away. For the Michelle Obama keynote, I was in Moscone West and the event staff was already redirecting people from the two overflow keynote viewing areas there to the Marriott. It was probably ten minutes after that when I heard them saying the Marriott was full. Needless to say, I found the best way to watch most of the keynote presentations was from my laptop in my hotel room. While I occasionally had drop outs (thanks Hilton wifi), not having to deal with the crowds was a huge advantage. At the same time, it was a little disappointing to know that, when they had a give-away based on a "golden ticket" at the end of the Sales Cloud keynote, I was fairly confident that my hotel desk chair didn't have anything taped underneath it.

For all of the crowds, it's also worth noting that there was a stepped up security presence at this year's conference. You could see security staff at any event you went to, any gathering of people. In all, while it was an ugly reminder of recent events in the world, it was some comfort seeing all of the efforts that were done to ensure everyone's security. In all, it was present, but unobtrusive.

The Expo
In some ways, you had to feel bad for the companies exhibiting at the Dreamforce Expo. This year, the Expo was in the reopened South hall. However, the current layout (and, from my understanding of Semicon West in July, this will also be true through that time) has the only entrance to the show floor down near the 3rd Street side of the hall -- essentially, the 100 - 400 aisles in an event. It made going into the South hall kind of like venturing into a cave, with the far, 4th Street side feeling rather disconnected from the entrance. To make things more accessible, Salesforce opened one of the emergency exits near the 4th Street end out to the Howard Street "Dream Valley" and decorated it with a bit of artwork suggesting you were exploring a cave.

Additionally, most of the sponsors were relegated to using a standard inline 10x10 system, so you had rows and rows of partners, all struggling to differentiate themselves or display some sort of unique identity. Mostly, as somebody walking through, it was like walking through endless white wall partitions. As a result, I found myself spending little time there.

Conclusion
In summary, it was another event and another year. Most of the time, I found little joy in being there. Would I recommend it? Hard to say. There are certainly some interesting take-aways, but at the same time, I'm not sure how well it stacks up against the raw costs in terms of ROI (when you consider the astronomical cost of hotels -- ~$395/night -- and the full $1500 conference cost). When you add in the oversubscribed crowds, you'll probably find that you're asking yourself why you subjected yourself to this -- an even more common thought than, is this worth it.

Monday, November 6, 2017

More on Facebook Manipulating the 2016 Election

As I wrote about back in 2014 in this post, Facebook Can (and has) Manipulated You With Their Feed Algorithm, we already know that Facebook can manipulate it's audience just by tweaking their algorithm regarding what's displayed. In much of the discussion regarding Russian advertising during the election and the various fake accounts that were being run by these Russian interests, I haven't seen this referenced once in the media.

And yet, after I wrote this post on online advertising the other day (and after I had a chance to see some of the ads and other Russian programs released by Congress), a broader realization occurred to me.

While the 2014 research by Facebook involved Facebook tweaking their algorithm, what if, instead of tweaking the algorithm, you simply changed the nature of many of the resources that the algorithm drew from? In other words, if the core algorithm behaved neutrally, but the assets that it presented were broadly colored, it would probably have the same effect as changing the algorithm.

Again, back to the point that I wrote in the previous post -- it doesn't really matter whether people clicked on Russian ads or not. Their presence there worked to change the color of their target audience's feed on a much broader scale.

Thursday, November 2, 2017

Underestimating the Impact of Facebook and Political Ads

I happened to catch this story on Talking Points Memo, Rosenstein: Americans Not ‘Influenced By Ads Posted By Foreign Governments’. In the post, here's the key quote from Deputy Attorney General Rosenstein.
“American citizens are pretty savvy, and they decide who to vote for,” he said in a podcast interview on the Target USA network. “I don’t think they’d be influenced by ads posted by foreign governments.”
Previously, I've also seen a similar quote from House Speaker Paul Ryan. The general theme is sort of a play to the vanity of an audience as a way to psychologically put a thumb on the scale of perception. Phrased a different way, it's sort of like, "come on, millions of people see infomercials every day, but people like you and I are smart enough not to actually order those products."

Of course, the obvious extension of that is that, some people do order these products. In the same way, it's easy to assume that no one would click on an attachment in a strange email or no one would fall for a Nigerian prince scam. But they do. In pure advertising terms, you can call these conversions. And while it's easy to say that well over a majority of people viewing ads, impressions, wouldn't convert, that's normal. With Adwords, Google considers 5% a good conversion rate. That means that 95% of the people who searched a specific term will see ads on that term without clicking on them. That's a successful ad.

The Impact May Have Been Much Greater Than A Conversion Rate Implies
The impact of these Facebook ads may have been far greater than the specific dollar amount sited or that any conversion rate implies. There are a couple of reasons behind this.

First, one aspect that hasn't really highlighted in the media coverage of these ads is the nature of the campaigns being run. For example, if the ads were running on a cost-per-click (CPC) model, then it's possible that the optimal cost result (for the people placing the ads) would have been few clicks, but many impressions. Essentially, this is like a brand identity campaign -- you don't necessarily need clicks, just the impressions can build mind share.

Another element might have used re-marketing campaign techniques to only show ads to people who had visited other specific content in the past. Using this technique, you could help identify and heavily target the ones who were most likely to buy into your message.

But perhaps the most important thing to keep in mind is related to this story from back in 2014, Facebook Can (and has) Manipulated You With Their Feed Algorithm. In summary, what Facebook posted a paper detailing how they modified the algorithm that delivers the feed to users, and for one set, they delivered more negative, less positive emotionally colored results, and for the other set, they delivered more positive, less negative emotionally colored results. As you might imagine, both groups essentially began echoing and amplifying the tone of their feed -- the negative group producing more negative content while the positive group produced more positive emotions.

As I wrote back in 2014,
For me, I see a much bigger danger implied by the Facebook study reference here. If you think of the ability to influence in this way, then the Facebook "scientific" paper is potentially a press release announcing their offering of an entirely new type of advertising. Imagine if you wanted to broadly shift public option. With the right amount of money and access to the right platform, you could pay for shifting the filter of the algorithm.  
What if you didn't need to broadly shift public opinion, you just needed a marginal push? What's more, you probably don't need to change the whole algorithm -- the algorithm already responds to a certain level of political polarization. Instead, perhaps all you need to do is color in some of the spaces...

In summary, I'd say, don't buy into the notion that these ads had no impact, that there's no there there.

Tuesday, October 24, 2017

YA Apple FAIL: iOS 11 Worse at Math that Intel Pentium

TL:DR: iOS 11 speeds the iPhone down the path from best phone ever to complete laughing-stock embarrassment.  

In case you were wondering, iOS 11 sucks, big time. Sucks. Sucks. Sucks. I've been intending to write a more comprehensive breakdown of the suckage, but other things have been keeping me too busy to devote the time to fully detail the flaws that I've found. It goes without saying that, if it were possible to go back a version -- other than any security flaws that they've theoretically patched, downgrading would be one hell of an upgrade.

I've posted previously about how for Apple iOS 11 and "No doesn't really mean no." because all of those services you think you shut off in the Notification Center don't actually turn off.

Additionally, if you've tried to use the notification center to select an audio source (like say, select you Apple Airpods), selection is now two clicks away under a tiny thumbnail of a control panel.

Should you want to change the brightness of your iPad while watching a video on Netflix, you're looking at two clicks or more to be able to make a simple change to screen brightness.

And now, I just came across this bug on MacRumors. iOS 11 Bug: Typing 1+2+3 Quickly in the Calculator App Won't Get You 6. That's right, if you type 1+2+3 quickly in the calculator app on your iPhone in iOS 11, the calculator will show you 24 instead of 6. In short, it's a worse math error than the infamous Pentium math error. The article goes on to detail an animation bug that blocks screen inputs while the animation resolves. And while it's probably a simple correction for Apple to fix their Calculator app, I think it's symptomatic of broader touch screen interface issues that have been introduced with iOS 11.

While using iOS 11, I've had numerous issues (somewhat randomly) with inputs to the touchscreen. Often, inputs near the edge of the display on my iPhone 5se fail. I've also had issues where the Messages app won't enable you to get to the bottom the display / see the most recent messages.

Perhaps the funniest thing about all of this was when I saw some pundit writing about facial recognition on the upcoming iPhone X, saying something to the effect that they were confident that it would be good because "Apple wouldn't let another Apple Maps - like fiasco take place." Yeah right. All I could think at the time was look at the joke that is iOS 11.

iOS 11 is so bad, it makes a compelling case for not updating your Mac OS to High Sierra.

Thursday, October 12, 2017

Facebook and the Problem of 'Bots'

Recently, there has been growing coverage in the media about Russian advertising (and influence) during the run up to the 2016 election. Some of this was driven by fake users (bots) on Facebook. While I'm thinking about writing a longer post about the election influence operation and online marketing in a broader context, one thing jumped out at me recently that I think is worth highlighting as it's somewhat misrepresented in the media.

In this USA Today article, the author wants Facebook to "Guarantee that bots will no longer be able to impersonate humans on the platform."

This fundamentally misunderstands the problem with "Bots". It's not like Facebook ever sat down and invited Bots onto the platform. Bots are carefully crafted bits of code that are scripted to mimic humans as you go through typical online activities. Bots are not obviously bots.

On one of the web sites I run, over the years I've had bots submitting inquiry forms thousands of times (Dear Salesforce.com: Web-to-Lead/Case Spam Sucks). Often, the form submission is some form of Spam. Even an inquiry form on a site in a niche industry can be a target for this type of activity. But what was actually interesting, in a way, was watching the form bots evolve. Essentially, when even when you make it more difficult for the bot to fill out the form, the bot kept exploring the parameters and requirements until you'd see it coming through again.

In that way, initally, my best defense against the form spam bots were to look for aspects that made them seem not human and try to filter against those. But eventually, you get to a point where if the bot fills out the form like a human would, you can't tell the different between an automated form engine and a human.

As you'll note in my Spam Post, at the time, Salesforce.com recommended that I install a Captcha, one of those image recognition test tools on the form. You know the tests, sometimes they're difficult to solve, even as a human. Sure, they provide an increased barrier for bot traffic, but they also provide a significant barrier to user engagement. Imagine if every time you wanted to post something on Facebook, you had to face a Captcha test?

And this is the fundamental problem with the "don't allow bots on your platform" arguement. It's just BS. Something being promoted by someone with a very simplistic view of the problem.

Friday, October 6, 2017

Apple iOS 11: "No" doesn't really mean "No"

Recently, Apple acknowledged that when you use the control center to turn off Bluetooth or Wifi in iOS 11, it doesn't actually turn those services "off". Instead it just disconnects from the things you were connected to -- except Apple devices and some Apple services. That's right. Apple's iOS 11 now features "No" doesn't mean "No" technology.

The Electronic Frontier Foundation warns about this in an article, iOS 11’s Misleading “Off-ish” Setting for Bluetooth and Wi-Fi is Bad for User Security, regarding this issue. They note that
When a phone is designed to behave in a way other than what the UI suggests, it results in both security and privacy problems. A user has no visual or textual clues to understand the device's behavior, which can result in a loss of trust in operating system designers to faithfully communicate what’s going on.
In gets worse.
The Wi-Fi will turn back full-on if you drive or walk to a new location. And both Wi-Fi and Bluetooth will turn back on at 5:00 AM. This is not clearly explained to users, nor left to them to choose, which makes security-aware users vulnerable as well.
Wifi and Bluetooth also reactivate when you reboot your phone.

Why did Apple build iOS 11 this way? Because Apple decided that it would be better for them -- make their device interworkings seem better -- than it would be to honor the "no" implied by the controls in the Notification Center.

This isn't the only thing that sucks about iOS 11, but it may be one of the more troubling ones.

Monday, October 2, 2017

Airpods: Apple's Imperfect Audio Port Solution

With the introduction of the iPhone 7, Apple decided to remove the classic audio port from the phone. They claimed to address the elimination of the audio port with a brand new product, Airpods, wireless Bluetooth headphones that would be so awesome, you'd never miss your old headphones. Skip the part about how Airpods weren't even available for the first three months or so of the iPhone 7. Oh, and the part about the limited availability of Airpods even today. Airpods were promised to be be so great, we'd never want to go back to wired headphones.

I have a set of Airpods, had them for several months, and I can only say, "Airpods, you are no wired headphones."

Don't get me wrong, Apple has accomplished something impressive with Airpods. They are, hands down, the best wireless iPhone headset that I've used to date. But that comes with a large number of caveats. But what's important to understand is that, while they're a nice iPhone accessory, they don't come close to replacing my earbuds.

Bluetooth < Wires
Fundamentally, Airpods suffer from the Bluetooth wireless connection. Using them in my office where there are numerous other Bluetooth devices, I experience a lot of drop-outs from both the speakers and the microphone. I blame this on Bluetooth interference. It's so bad that if I have a call, I use my wired earbuds.

But the drop-out problems aren't limited to "Bluetooth noisy" environments. Using the Airpods while listening to music walking around at night, I found the sound dropped several times also. Potentially, this could be related to software that has each ear piece to check in with the other one and verify that it's there -- the auto-detect ear software. Apparently, you can turn this off, but I think I saw where disabling this disables stereo -- which makes sense if you consider that it needs to know whether there is a second headphone in order to send a stereo signal. However, it should be noted that my wired headphones don't experience similar drops.

This brings me to the sensors in the Airpods. With most Bluetooth headsets, there is some functionality that enables you to answer calls with a tap -- similar to hitting the start-stop button on the wired headphones. Unfortunately, with several months of use, all I've succeeded in doing with my Airpods is opening the voice interface at various times. I'm sure that if I were running battery-wasting, time-wasting Siri, I could have asked Siri to answer the phone, but mostly I've had to default to scrambling around, trying to find my phone in order to answer calls on the iPhone instead of the Airpods.

On a related note, I can't count how many times I've accidentally fumbled with one of the Airpod earpieces, only to accidentally bring up the voice interface -- enough to make me cautious about handling the earpieces.

The Good
Let me say that I understand the interest in having a wireless headset. How many times have I found myself listening to music while doing something like cooking when the wire on my headphones gets caught on a knob or the corner of a cabinet? Having my ear yanked or sending my iPhone flying sends me cursing the wire and everything it stands for. With the Airpods in a quiet Bluetooth environment like my home, I can wander around with audio, safe from worrying about whether the cable is about to get caught on something. Do I still experience drop-outs? Yup. But the trade-off of no wires is usually worth it for kitchen tasks.

Airpods also work reasonably well when driving in my non-Bluetooth-enabled vehicles. The wire doesn't usually get in the way while driving, but it's still potentially a concern. Meanwhile, the variability of cell coverage when you're on the road means that the wireless connection is only one of multiple connectivity issues that you deal with.

Finally, another nice thing about the Airpods is that you can use one or both. If you use just one, you can get some pretty serious battery life out of the set by periodically switching earpieces and letting the other one charge.

Summary
All trade-offs aside, Apple's Airpods provide a nice Bluetooth headset experience. At the same time,  one of the reasons that I chose the iPhone 5se was the inclusion of the headset port. Having experienced both the "antique" and what Apple promises us to be the "future" of audio on a phone, I must say that the removal of the audio port still ranks as one of the worst design choices Apple has made.