William Meisel's Blog

Current economic news validates the concerns in The Software Society about the impact of automation eliminating jobs too rapidly for the economy to properly adjust. The strength of the recovery is brought into question by Labor Department revisions last week reducing the reported number of jobs created in June and July by a total of 74,000 jobs. And statistics continue to affirm that the jobs being created are on average at lower salaries than the jobs the people taking the jobs were previously earning.

In another key statistic, the drop in the unemployment rate is largely caused by people leaving the workforce rather than the creation of new jobs. If most of those people are taking early retirement, then they will generally have less to spend and put an increasing burden on pension plans and Social Security.

The causes of the slow pace of recovery usually cited are the equivalent of “weakness in the economy,” as if the subprime crisis was the sole cause of current problems and we just have to wait until its effects wear off. I argued in The Software Society that the core issue today is technology acceleration driven in part by digital systems and software—rapid changes driving efficiency, but at the cost of jobs. Increased productivity has usually benefited the overall economy in the past, even when some industries suffered from Joseph Schumpeter’s “creative destruction.” However, current technology changes may now be occurring too fast for companies and individuals to adjust effectively. The issue of automation proceeding too quickly (what I’ve called “over-automation”) was a major topic in The Software Society and in other books such as Race Against the Machine by Brynjolfsson and McAfee and The End of Work by Jeremy Rifkin. A tendency in the press to emphasize what good news there is—small improvements in statistics—ignores there being 1.9 fewer jobs than at the economic peak in 2008.

Some authors have suggested that an approach to the problem of over-automation—too few jobs as a continuing rather than temporary problem—is essentially socialism. If there aren’t enough jobs, the argument goes, the government can simply take more income from companies or even run them and distribute the income to individuals. Hopefully, history has proven that this isn’t a real alternative; governments aren’t very good at directly or indirectly running companies, with results ranging from inefficiency to pollution to corruption.

I suggested an “automation tax” in The Software Society as an approach that pushes companies in the right direction without requiring any specific actions. The tax could be structured as tax credits and penalties that depended on companies choosing ways to increase productivity that made employees more effective rather than replacing them. If the net effect of the tax was increased government tax revenues, those revenues could be applied to covering the increasing burden of Social Security, a benefit that many people working in the US have contributed to with the expectation that they could count on it.

There are some compensating trends in the US that provide time to address the problem of over-automation. One is that new, highly automated factories are bringing some manufacturing back to the US, a positive, but transitory, impact of jobs being automated away. Another is shale oil/gas production in the US, which creates new jobs, but has obvious limitations to its long-term growth. The positive aspects of the US economy may give it time to lead an international recovery if we address the core problem of over-automation.

Many readers of this blog have a technical background and may feel insulated from this trend to the extent that their companies are selling automation solutions. And technology that makes humans more effective at their jobs can indeed contribute to preserving rather than replacing jobs.

Ultimately, however, there must be buyers at the end of the sales chain. If there are no jobs, there is no income to buy products, no matter how efficiently they are produced. Government policy should motivate companies to act in service of the overall economy rather than favoring short-term profits.

In The Software Society and in this blog (Where are the jobs?), I’ve argued that celebrating each quarter that we’ve added a few jobs misses the issue, noting that the quality of jobs and median income have continued to decline. I argued that “productivity” has evolved into “over-automation,” a preference by companies to “hire” computers rather than people. Increasingly, the jobs being eliminated by technology impact the economically and socially critical middle class at a rate faster than the workforce can adapt. Economist Joseph Schumpeter’s “creative destruction” caused by technology advances now is doing more destruction than creation.

The Wall Street Journal on August 10 discussed some statistics that emphasize this trend in “Roots of the Living Wage Wave.” The article said, “Economic changes of the past decade have led to a decline across the country in well-paying jobs, such as those in manufacturing, and an increase in jobs that pay less, such as those in hotels and food services…Positions are increasingly being filled not with the young and inexperienced, but older and more skilled workers who can’t find other jobs.” The decline in average buying power implied by this trend must eventually lead to an overall decline in sales and profits for all companies. Lower wages today may temporarily boost profits, but the long-term macroeconomic effect will be a decline in GNP. A company can’t maintain revenues without buyers.

Unfortunately, the “living-wage” solution discussed in the article is a trend toward local governments requiring a higher minimum wage; the Washington DC city council, for example, has taken such a step. By raising the minimum wage, one could argue that low-paying jobs can simply be turned into higher-paying jobs. The result in DC is the usual unintended consequences when a government tries to set prices or wages. For example, Walgreens is threatening to open fewer stores (and thus create fewer jobs) in the city.

The better solution is to restore classical middle-class jobs so that there is less competition for the lower-paying jobs, and so that employers must pay more than the minimum wage to get good employees for those jobs, letting Adam Smith’s “invisible hand” create a balance. Government policies must encourage using technology to give employees better tools rather than using technology to replace the employees. Today’s government policies promote the opposite—preference for machines over people through costs and obligations for employees such as payroll taxes and health insurance while driving interest rates to lows that make capital investments in machines cheaper.

In The Software Society, I outlined several solutions that would help tip the scale toward the human solution, including an approach based on the corporate tax system. This isn’t the place to argue the details, but the core problem of over-automation must be addressed if there is to be a healthy long-term solution to today’s worldwide economic malaise.

TV is evolving to allow watching What You Want When You Want (let’s call it “WYW2”). For example, Google’s inexpensive Chromecast is a device that plugs into compatible TV sets and lets you stream content to the TV through an app on a smartphone as well as asking for viewing options by voice. Comcast Cable’s Xfinity TV, launched in Minneapolis-Saint Paul, includes a new X1 remote control app for an Apple iPhone that allows customers to find content through motion, gesture, and voice control.

WYW2 adds to conventional TV choices the digital content we can get on our PCs and mobile devices. The extensive content motivates voice interaction in “natural language” to find shows (just say What You Want), and voice control is explicitly included in most announced “next-gen” TV plans.

The option to interact easily with a TV (as opposed to the limitations of classical remote controls) raises an interesting possibility. Does it create a new category of programming that blurs the boundary between TV and games? The distinction in the past between games and TV programs seems to be that TV viewing is passive–essentially no interaction–and games are intensely interactive. Is there something between these extremes that voice interaction with TVs makes possible? Something that creates new opportunities for creative content?

I believe so. I suggested in The Software Society that the maturing of speech understanding allowed such an intermediate model, one with moderate interaction with a viewer that impacts what is essentially a video show. An example I gave was a murder mystery where the viewer is the detective and “interviews” suspects in some scenes to find the villain.

A well-designed interactive show might even encourage multiple viewings to see what happens when the viewer responds in a different way. An ad that viewers might watch repeatedly to see alternative responses would certainly be an enticing concept to ad agencies. A “help assistant” that shows you videos on how to use a device, software package, or service–while allowing you to ask questions and influence what is shown–is another attractive option. The use of a near-talking remote (e.g., a smartphone) and advanced speech recognition in the network make complex and natural interaction technically feasible, as today’s personal assistants on smartphones demonstrate.

The current trend in TV control enables new alternatives between passive watching and game-like intensity. What You Want is a moving target.

Recent financial reports by consumer technology companies have disappointed analysts. The general consensus seems to be that the cause is a lack of new products and innovation. The underlying cause may be more fundamental.

We may have reached a plateau in consumer’s ability to absorb the increasingly rapid change in technology. Consumers are becoming saturated. They are finding that new products and features don’t add much to their experience with technology, and sometimes simply constitute a challenge in adapting to a modified user interface. The core cause of saturation is limitations on our human abilities. Those aren’t likely to be upgraded soon.

For example, most PC users find their PCs are fast enough; the real limitations on efficiency of use are the user’s typing and thinking speed. Features such as better screen resolution are limited by what our eye can actually see.

Today’s smartphones and tablets have most of the features we need and can handle. The limits of usage are often the fact that there are only 24 hours in a day, another unchangeable fact.

New features and products must contribute substantially to efficiency and ease of use to make a real impact. They won’t succeed if they do the opposite, slowing use of the devices because of non-intuitive behavior or a steep learning curve.

There is room for growth in some technology areas not fully developed, e.g., making it easier to manage TV viewing as it evolves to “watch what you want when you want to watch it.” But humans aren’t changing as fast as technology (a major theme in The Software Society), and a day will remain at 24 hours. There is a saturation point, and we may be approaching it.

Amazon’s pricing for The Software Society:

ebook: $3.99 (standard), hardcover $6.33 (temporary!).

Microsoft’s recent reorganization focused largely on unifying its products and services. The reorganization is a symptom of a major economic trend that I call “Technology Communities” in The Software Society. Briefly, technology communities are based on families of products and services that work well together. If a buyer has a product from an effective community, the buyer is more likely to buy another or an update from the same community. There is usually a single company that is a de facto manager of a community and drives innovation in that community.

Apple is the clearest example of such a community. Apple products share many common features and work well together. The iPhone and iPad have the same operating system. New photos from an iPhone are automatically downloaded by iPhoto on a Mac when the phone is connected. iCloud can connect everything, so your iPhone warns you when you have an appointment even though it was scheduled on your Mac. The list goes on, with Apple constantly trying to add to it, including encouraging car manufacturers to build in a connection to the iPhone’s Siri. You might eventually set your bedroom alarm clock and your household thermostat with Siri.

Another major community is driven by Google, unified by the Android operating system and the pervasiveness of Google’s web search and maps, even on Apple devices. Google has been losing some control of this community; Companies such as Samsung overlay the Android OS with specialized features such as the S-Voice personal assistant, a connection to the customer that is beyond Google’s control. Google’s acquisition of Motorola Mobility and expansion of personal-assistant-like natural-language voice search are in part a response to this threat to their control of the community.

Microsoft still has a strong Windows/Office software community in PCs, but is losing a growth opportunity as communities expand to multiple devices such as smartphones and tablets. Windows 8 is an ambitious attempt to unify the Windows community across devices. Office 365 is an attempt to leverage Microsoft’s dominance in productivity software in the mobile world. The reorganization is an attempt to unify these and other efforts.

In part, the growth of communities is driven by what I’ve called in this blog “digital overload”—Too many devices, too many features, too much communication. Yes, we want many of these devices, but we’d like using a new one to be based on what we know, and not a constant challenge. The growing dominance of technology communities is a response to our reaching a saturation level with “new.” Companies assume younger members of society always want a major innovation, a position that is probably not true for the majority of that group and ignores an aging population.

In an attempt to make the next device a true “breakthrough,” companies are adding features such as gesture control that are more of a gimmick and a battery drain than a true addition to usability. A focus on always adding complexity misinterprets history. The iPhone adopted the familiar Graphical User Interface from PCs, simply using a finger as the pointing device rather than a mouse. The main applications were familiar and didn’t demand a learning curve, e.g., email and web browsing. The iPhone was an innovation, but one that heavily leveraged an established user interface paradigm. The iPad took the smartphone interface and put it on a larger device that made some features such as watching video or reading books easier, an evolution easy to master, particularly if one already had learned iOS on an iPhone.

The core impact of technology communities and feature saturation on the business of technology is not fully understood by most companies and by the business press. For example, there has been much discussion of the declining sales of PCs, usually assuming people are replacing them with tablets. The true cause is that the major historical motivation for upgrading was the core features of more speed and more memory, and most of us now have more than enough speed and memory for what we do on PCs. The future motivation for upgrading may be tighter integration with a community, e.g., built-in dual directional microphones with noise cancelling to allow us to easily speak to our ubiquitous personal assistant when using our laptop. A full exploration of technology communities and their implications is beyond the scope of this blog entry, but companies must recognize the trend and make it part of their strategic planning.

On June 25, the Division of Advertising Practices of the Federal Trade Commission (FTC) sent a letter to search engine companies to update guidance published in 2002 on distinguishing paid search results and other forms of advertising from search results based on objective criteria. The updated guidance is directed in part at smaller devices such as mobile phones where the distinction of ads by shading and labeling may not be as clear as on a PC (e.g., if only the advertised options are visible on the small screen). The letter also explicitly says: “if a voice interface is used to deliver search results, a search engine should make an audio disclosure that is of an adequate volume and cadence for ordinary listeners to hear and comprehend it.” The letter also specifically notes “the introduction of voice assistants on mobile devices have offered consumers new ways of getting information,” and cautions, “Regardless of the precise form search may take in the future, the long-standing principle of making advertising distinguishable from natural results will remain applicable.”

The letter notes, “…consumers ordinarily expect that natural search results are included and ranked based on relevance to a search query, not based on payment from a third party. Including or ranking a search result in whole or in part based on payment is a form of advertising. To avoid the potential for deception, consumers should be able to easily distinguish a natural search result from advertising that a search engine delivers.” The letter also cites an indirect form of paid search: “Sometimes the results returned as part of a specialized search [‘such as news, images, local businesses, or consumer goods’] are based at least in part on payments from a third party. If that is the case, it is also a form of advertising and should be identified as such to consumers.” (This caution seems to include cases where the source of the information, e.g., restaurant reviews, is a preferred provider that pays for that role.) The letter cites court cases that have concluded that disclosures must be “sufficiently prominent and unambiguous to change the apparent meaning of the claims and to leave an accurate impression,” and that “simple, unequivocal” disclosures that are “conspicuously and prominently made” are required to overcome any misleading impressions created. The FTC published more detailed guidelines in a March update to the 2002 document.

I argued in The Software Society (and repeated in this blog) that that a major trend in user interfaces was the “Personal Assistant Model” (PAM) with a major factor being that the assistant used natural language processing to understand a request and deliver an answer as directly as possible. The PAM providing a direct answer rather than a list of options was particularly important in voice-only mode, e.g., while driving or using a Bluetooth earpiece microphone. A wearable device such as Apple’s rumored iWatch would necessarily have to provide concise search results as well. Listening to a long list of alternatives, some loudly stated as “ads,” is not an attractive alternative.

Despite the FTC’s concerns, I think everyone understands that advertising supports Web search providers, and no one wants to do without web search. Personal assistants such as Apple’s Siri today aren’t as obvious a source for advertising revenue, but in fact are the future of search. One Google executive has testified to Congress that Siri is a threat to their ad revenues, and another has stated at a conference that the trend to give more direct answers rather than solely a list of web sites is the “end of search as we know it.” Siri will launch a search (now Microsoft’s Bing), but also bypasses a general search to directly answer inquiries such as “Did the Giants win?” or “What movies are playing?” Google’s expanding collection of search features, including its Knowledge Graph, Google Now, and voice search accepting natural language, come close to a PAM. Repetitive statements that the source of the info or the choice of a specific result is an “ad” certainly will be annoying and inefficient, particularly if the message is audio and can’t be ignored.

It is critical to the usability of search and to the PAM that the FTC allow some minimally obnoxious way to indicate that a result is influenced by payment. Perhaps there should be simply a tone (“cash register”) before an audio result impacted by payment, with the meaning of that tone being advertised or indicated clearly once when a search engine or personal assistant is engaged. A repetitive statement before every reply that “this is an ad” is a bit like a pop-up ad on a web page and would be equally irritating.

There should also be a standardized option to get alternatives not favored by payment, which might minimize FTC concerns. A button at the top of every web page that allowed hiding advertised options might be sufficient for visual displays. For voice commands, a standard phrase such as “no ads” after initial results are displayed could be an option that doesn’t make it too easy to avoid ads.

The trickiest case is where there is a preferred provider of information, e.g., Yelp for restaurant reviews or a specific news source. In some cases, the personal assistant provider gets some fee for directing a user to the specific web source. Calling this an “ad” can be misleading, for example, when delivering factual information such as a sports score, even though the source of the information may be a preferred provider. I believe it is sufficient in these cases for the personal assistant provider to list preferred providers in the terms and conditions that users accept to use the personal assistant service. A standard voice or text command such as “preferred providers” could display a current list.

In any case, the FTC’s guidelines shouldn’t make the user interface a nightmare. That’s in the public interest as well.

The Federal Reserve recently made comments suggesting it is cautiously optimistic about the economy and would begin reducing its bond-buying program when certain goals are met. The Wall Street Journal in an article “Slow-Motion U.S. Recovery Searches for Second Gear,” June 24, 2013, authored by Brenda Cronin, sounded a similar tone of cautious optimism.

The caution was reflected in cited statistics: “The U.S. has 2.4 million fewer jobs today than when the recession began. Adjusting for population growth, it will take more than nine years at the current rate of hiring to return to prerecession employment levels, according to estimates from the Brookings Institution.” Other statistics listed in the article revealed that business investment in equipment and software rose by 35.2% and industrial production (manufacturing) rose by 20.2% during the recovery starting in 2009, but private industry jobs rose only 5.4%, a disconcerting mismatch.

These statistics support the concern expressed in The Software Society and an earlier posting in this blog that industry is choosing to replace jobs through automation rather than hiring workers. Statistics not cited in the article suggest that even the few jobs being created represent workers taking jobs at lower salaries. The Bureau of Labor Statistics reported on June 5 that, in the first quarter, while nonfarm business sector labor productivity increased at a 0.5% annual rate, there was a 3.8% decrease in hourly compensation. The report noted, “The decline in hourly compensation is the largest in the series, which begins in 1947.” Median income has also dropped more than 5% during the “recovery.”

Government policies indirectly encourage “hiring computers” over hiring people. The low interest rates artificially created by the Federal Reserve make it inexpensive to borrow to finance automation. Government requirements such as payroll taxes and requiring health care insurance increase the cost of employees. It seems rational for individual companies to minimize headcount, but, as dropping income reduces consumption, the economy will eventually contract, affecting all companies.

Over-automation is reminiscent of the subprime mortgage debacle, when all the banks assumed that they had to make poor loans because all their competitors were doing it. The danger in most companies embracing over-automation is perhaps less obvious, since “productivity” has historically been good for the economy. But, nevertheless, there is no rational argument that the process can’t go too far. In their 2012 book, Race Against the Machine: How the Digital Revolution is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy, MIT professor Erik Brynjolfsson and Andrew McAfee highlight this issue.

Government action is required to break the downward spiral driven by computers eliminating jobs. In The Software Society, I suggested an “automation tax” to counter government policies that indirectly encourage hiring computers rather than people. Others will have alternative solutions, but we must admit the problem before there can be serious discussion.

A University of Utah study released by the AAA Foundation for Traffic Safety on June 12 was widely reported as concluding that creating text messages by speech was dangerous. The study concluded that texting a friend verbally while behind the wheel caused a “large” amount of mental distraction compared with “moderate/significant” distraction for holding a phone conversation or talking with a passenger and with “small” distraction when listening to music or an audio book. I don’t dispute the findings as stated, but the comparison itself is flawed if used to conclude there is no benefit to creating text messages by voice when driving.

If you are going to text someone while driving, then you are making a decision to compose a message to be delivered while driving. The real question is, once that decision is made, is texting by voice safer than texting by touch keyboard? The answer should be obvious, unless you can type on a mobile phone with one hand without taking your eyes off the road, even if you assume it is equally safe to drive with one hand versus two on the wheel.

Texting is more difficult than speaking or simply listening. You are composing a message intended to be read rather than heard, and you want it to be accurate and clear. The fact that a text message must be short increases the cognitive load as one must think how to deliver a message as briefly as possible. This takes a different level of mental involvement than conversations where a passenger can engage you in quick dialog to clarify a statement, for example.

The difference will be clear to those of us who have used dictation to compose documents. It’s not like talking to a person. You are speaking something to be read as text, which has a high expectation of clarity. Dictation requires effort comparable to composing a document while typing. Dictation of a text message without looking at the display in fact probably requires more concentration than typing a document when you can see the results on a PC screen; when working eyes-free, you don’t see what you have written and must maintain context in your head.

Composing text is clearly more of a mental load than conversing or listening. Concluding that this is the case with a study is not particularly revealing. It has nothing to do with whether, once you have decided to text while driving because it can’t wait, doing it by voice is safer.

I blogged before Apple’s World Wide Developers conference that I believed Apple was well-positioned to address the problem of  “digital overload,” the expansion of digital devices, features, and services that is beginning to overwhelm users rather than make their lives easier. Apple, with its range of devices, could unify the experience across devices, making using the different devices similar enough to avoid a new learning experience with each device. For example they could make an expanded and improved Siri a “ubiquitous personal assistant” available across platforms. [See my blog entries Ubiquitous Personal Assistants and The Ubiquitous Personal Assistant: The battle has begun or the deeper discussion in The Software Society.]

Instead, Apple choose to emphasize a visual redesign of its Macintosh and iOS operating systems. The visual redesign simply changed the look of the product in ways—to my eyes—that were more distracting than constructive, e.g., a translucent page that showed a fuzzy image of the page it overlaid. Icons look different, but does that significantly change the user experience? And providing the option of a male voice for Siri doesn’t change its utility.

The features that were added, initiated by various touch motions, run the risk of adding to digital overload. It is too easy to invoke an action one didn’t intend by inadvertent touches or motions, and then find oneself struggling to undo an unwanted action.

Apple did make some small moves toward reducing digital overload. The iCloud “keychain” which unifies passwords across devices and services addresses a common digital-overload issue. The company’s Pages, Numbers, and Keynote are productivity tools that work in a browser and thus across devices (even a Windows PC). Cross-device features help, e.g., allowing a navigation map for a specific destination created on a Macintosh PC to be sent to an iPhone easily. The company has integrated one additional platform, the automobile, with an initiative that places an iOS screen in automobiles, supported by Siri.

Siri can now control more iPhone functions, including turning on Bluetooth or increasing screen brightness, and the personal assistant can answer a wider range of questions. Adding things Siri can do with a simple, intuitive voice command can reduce the burden of learning how to use such features (or even discovering if they are there), but Apple made those features a minor part of their presentation.

Overall, some items in the announcement do move a bit toward cross-device and natural-language control of more features that will incrementally reduce digital overload. It’s a shame that Apple chose to emphasize visual candy over real nutrition.

Apple has excelled at identifying an unmet need and serving it–the iPhone and iPad being obvious examples. To understand what Apple may do next, one should start with identifying a major problem with our digital devices and services that the company is positioned to address.

What is that problem? The problem is that we have too many devices, all with too many features and services. They all seem to demand attention, and they all have features we’ve not even discovered. Apple is in a strong position to address this problem of “digital overload.”

What is digital overload? Our digital systems may include a PC, smartphone, tablet computer, TV/digital-recorder, and automobile infotainment system. And, with the advent of Google Glasses, a wearable device is probably next. Features multiply with each new model and upgrade. It is a time-consuming and frustrating challenge to deal with all that complexity. The unmet need that Apple can address is to simplify the digital experience by making it more intuitive and consistent across devices. Make all our devices feel like one digital service!

A historical perspective is useful in understanding this opportunity. The Graphical User Interface (GUI)–a pointing device, windows, menus, icons–brought computing to the masses. The GUI’s familiarity helped us understand quickly how to use a smartphone or tablet computer. But the GUI is becoming over-burdened. For example, a menu is less efficient if it displays ten items rather than five, particularly if many of those items lead only to another long set of options from which to choose.

We also find an excess of features and options in each device. And we are bombarded with messages of various sorts. Proactive features such as Google Now have their points, but also can add to the constant digital demands for our attention.

So how could Apple address this need? It starts with improvements in Siri. The core breakthrough that will propel the next cycle of growth is natural language understanding and speech recognition technology. Apple’s Siri showed the potential of being able to ask for something the way you might ask an assistant. The objective as discussed in my book The Software Society is “Personal Assistant Model,” an extension of what Siri does now. The Personal Assistant Model is the equivalent of the GUI in terms of its potential to drive another cycle of technology growth. The key is that speech and natural language interpretation technology have reached the point where it can support this model at an acceptable level, and the technology will continue to get better.

How must Siri be improved to meet this goal? First, the core technology supporting Siri must be continually improved, both the speech recognition and the understanding of a request. Both are driven by analysis of data, and Siri has been collecting data on what we say to it in huge quantities. As a research manager deeply involved in speech technology told me recently—”it’s all about the data.”

Second, Siri must allow text input as well as speech, allowing us to type what we would say. That makes Siri usable when speaking isn’t a good option.

Third, Siri should work across all our devices. This is the key feature, making a unified interface that is the same no matter what digital system we are using. Siri can be ubiquitous, remembering what you did on one device when you are using another. For example, you should be able to tell Siri to text a message from your phone while at your tablet computer, using the contact list on your phone. The Software Society and my blog entries (Ubiquitous Personal Assistants and The Ubiquitous Personal Assistant: The battle has begun) discuss the Ubiquitous Personal Assistant in more detail.

Siri could also add features that reduce the aspect of digital overload that comes from too much communication and too much news? Ideally, we could simply say, “Show me emails from people I have replied to in the past” (or from a priority list you compile). Or “Show me relevant news,” where “relevance” is determined from what articles you have perused for significant amounts of time in the past. Such features require work to develop, but not any deep technology breakthroughs. It is easier for a company like Apple that develops the products that carry its software. Apple also has iCloud to help provide cross-device consistency. And Apple supplies many of the core applications on its devices, e.g., a calendar app on the iPhone, so that Siri can engage device-specific apps.

How will the Personal Assistant Model drive a unified strategy that could be part of a single major announcement? What will Apple do in TV, for example? The remote control for a TV could be an iPhone with a special app using a WiFi and/or Internet connection to the TV. The technical challenge of dealing with speech across a noisy room addressed to a TV or peripheral device will make it much more preferable to simply raise the iPhone near one’s mouth and ask Siri to “change to ABC News.” We could even tell Siri to have the TV record a program while away from the TV.

What about wearable computing? Apple CEO Tim Cook seemed to promote the idea of a wearable device at the D11 conference recently, pointing to a Nike wristband he was wearing. But he also said, commenting on Google Glasses, “There’s nothing that’s going to convince a kid who has never worn glasses or a band or a watch to wear one, or at least I haven’t seen it.” Whatever Apple’s concept for a wearable device (an iClip?), power and size considerations are likely to make it a Bluetooth-connected accessory for an iPhone, rather than a standalone device. And the small size would make voice interaction a requirement.

In the automobile, we will simply want the vehicle system to allow us connection to the smartphone we have already mastered, with a supplementary personal assistant for vehicle-specific services. Our personal assistant will have a voice-only option, as Siri currently does.

And Apple can engage outside developers through its App Store for specialized features. Optional third-party specialized personal assistants could register with Siri so she can engage them when asked.

In summary, the next big thing won’t be a single device or incremental features. It will be unification and expansion of Siri to a major feature on all future Apple devices and services. Apple will address the core problem of too many devices and too many features. Featuring Siri as the primary interface makes the introduction of new devices more seamless.

I don’t expect Apple to announce all this at the upcoming Apple Worldwide Developers Conference. I hope they will announce that Siri will accept partners in specialized personal assistants available in the App store, applications that, once downloaded, Siri can bring up by request.

Perhaps I’m just imagining what I’d like Apple to do. If Apple doesn’t unify our digital experiences, I hope someone else will. Google tries to be a device-agnostic option and has recently expanded search (including natural-language voice search). Microsoft has the core technology capabilities to move in this direction, already offering some voice search options in Bing. Nuance Communications’ CEO Paul Ricci recently expressed his strong belief in the ubiquitous personal assistant at a conference; Nuance provides an option for device manufacturers, and is already teaming with Samsung with its S-Voice personal assistant on Galaxy smartphones.

Test this proposition on yourself. Wouldn’t you like a unified cross-device experience that let you simply say what you want?