In this reprint from the October 2006 issue of Game Developer magazine, Michiru Yamane opens up about her musical education and her work on a few classic Castlevania games. She would go on to leave Konami in 2008, striking out as an independent composer and scoring games like Skullgirls.

Michiru Yamane is the principal composer for the game series Castlevania, among others. Her classical and dark arrangements in Castlevania: Symphony of the Night have earned her more prominence among game music writers, Konami fans, and the industry at large.

Having worked in games since the late 1980s, Yamane is one of the most recognized composers in the field. Game Developer spoke with her about her influences and inspirations, as well as her plans for the future.

How did you first start at Konami?

Michiru Yamane: Just before my fourth year of college, I started looking around in the recruitment office for different jobs I could apply to, and I found Konami there. That’s the whole story!

It seems like a lot of people I’ve talked to at Konami joined up right after college.

MY: I actually had a teaching license at the time, and I did get another job, but it was part-time, and it didn’t suit me at all. The school I was working for didn’t really like me that much either, so I started looking for another job, and that’s when I decided to join Konami.

Did you study music in school?

MY: As a kid I started learning piano, and I went to a high school that had an emphasis on advanced piano. It was specifically a musical high school. But there were so many piano virtuosos who had technical skill and I didn’t want to compete in that way, so I chose a university that had strong music composition courses.

What made you interested in this sort of music, this sort of old, gothic, Victorian style?

MY: I guess it comes down to my schooling again. When I was at university, my thesis was based on the music of Bach, so I was immersed with pretty classic yet dark music. But I don’t want to have that stereotype, if possible.

What was the first game you did music for?

MY: Ganbare Goemon 2 in Japan, then some arcade titles, and some Game Boy work, but often I only did partial soundtracks. I moved on to some Super Nintendo and Genesis stuff as well, with Castlevania: Bloodlines and Rocket Knight Adventures. I’d say my big break was Castlevania: Symphony of the Night.

What sound chip did you prefer from the old days?

MY: Probably the Genesis. For that system I got to do all the work, not just composition, but also synthesizing the music into the program. So it’s particularly memorable for me.

Do you ever have any influence on the design of games?

MY: Not from the early stages, but I can certainly influence design from a sound standpoint in terms of how things are implemented and the direction of events.

I remember a specific time when my vision and [Castlevania director] Koji Igarashi’s vision didn’t overlap. At that time, I basically had to change my way of thinking to match his, so maybe I don’t have that much control. But usually we’re on the same page.

With Castlevania specifically there’s a lot of revision of classic music, it seems. Is that a problem for you?

MY: I do get to create lots of new music for the series, even with arrangements. I did one for Symphony of the Night and several more for the Game Boy Advance. I really like the music of Castlevania already, so it doesn’t bother me at all.

What games have influenced you in terms of sound?

MY: I really love the Tomb Raider games — well, 1 and 2 anyway. They don’t have any music, but they have really good sound work, so that made me think more carefully about the way I use music in games.

It seems like lately the Castlevania series is becoming less and less popular in Japan. Have you considered hiring a Visual Kei band [a Japanese music style that integrates rock and classical, in a manner similar to Castlevania’s music] to do a tie-in with a future game?

MY: I haven’t, but it seems like a good idea. For the next game we do plan to integrate some music with vocals, but it won’t be rock. It’ll be more operatic.

What kind of music do you listen to?

MY: In the rock category, Dream Theater. I’d love to collaborate with them some day, but somehow I don’t think that will ever happen.

Do you ever make music just for yourself?

MY: Well, not really, unfortunately. Since I’m a Konami employee, I’m always devoting my full efforts to the music of whatever game I’m working on, usually Castlevania. But someday I’d like to.

I’m afraid that if I did make my own music, it would sound quite a lot like Castlevania music anyway…would that be all right?

Definitely!

You can find more great stories from old issues of Game Developer Magazine in the GDMag Archive section of the GDC Vault.

The ongoing global pandemic is affecting video game companies in different ways, but one notable shift highlighted in several recent quarterly financial reports is an increased interest in digital game sales.

Embracer Group, the parent company of THQ Nordic, Coffee Stain, and others, is no exception. The Group closed out the quarter ending March 31, 2020 of its 2019-20 financial year with 78 percent of its game sales coming from digital sources, a record high proportion for the company.

CEO Lars Wingefors suspects the split, and a recent “overall increased consumption of games” is largely due to the COVID-19 pandemic. It’s something he says is made more evident by the fact that Embracer had no major releases during March and that its already seeing a “boost in consumer activity” alongside game launches in the current quarter.

For Q4, net sales in the Group’s Games division decreased to SEK 903.5 million (~$94.2 million) from last year’s SEK 1.03 billion (~$107.8 million), though Embracer notes the decrease is exacerbated by last quarter’s high sales from the launch of Metro Exodus.

Its Partner Publishing/Film business saw net sales fall to SEK 435.6 million (~$45.4 million) from last years 595.6 million (~$62.1 million), a drop driven by the pandemic-driven closures of many physical retailers.

Despite those decreases, Wingefors calls Q4 a stable quarter and says overall profitability was right in line with its expectations, noting as well that Q4 made for the games division’s second-best quarter ever in terms of net sales.

For the full year ending March 31, 2020, Embracer saw net sales in its games business increase 31 percent to SEK 3,196.5 billion (~$333.3 million) while net sales for the company as a whole increased 3 percent to SEK 5.249 billion (~$547.5 million).

© Nintendo Life

Animal Crossing: New Horizons is something of a phenomenon, in case you hadn’t noticed. Not only has it sold by the bucketload, but it has also attracted celebrity fans like Danny Trejo and Elijah Wood, and seems to be the perfect tonic for the current coronavirus lockdown. Heck, we awarded it a well-deserved 10/10 when we reviewed it at launch, and in general, our admiration for this unique life sim has only grown over time.

However, any video game is likely to grow stale over time, even the best ones in the world – and now we’re a few weeks past the game’s release, it seems a fine time to ask that most vital of questions: are you fed up of Animal Crossing yet?

Damien McFerran, Editorial Director

Before I answer that question, it’s probably worth me outlining my own personal history with the Animal Crossing series, because I think it has some bearing on where I am right now.

I got the GameCube version thanks to the overwhelming hype which surrounded its release, and pretty much bounced right off it. I didn’t have the time to dedicate to such a massive undertaking, and therefore my first ‘real’ experience of Animal Crossing was Wild World on the DS – a platform which I could take with me anywhere and was, therefore, a better fit for my hectic lifestyle.

I immersed myself in Wild World’s charms, playing it solidly during lunchtimes and in the evenings. Then I kind of stopped. The allure passed pretty quickly and I moved on to other DS games. Next up was New Leaf, which was pretty much the same story – I went in hot, I thought the game was amazing, then put it aside as other, more pressing titles appeared.

Perhaps the key issue here isn’t with the game – which offers a staggering amount of gameplay potential and content – but with me?

I’m sorry to say that the story remains the same with New Horizons. For the first few weeks, I played it every single day without fail – not booting up the game meant I risked my island being overtaken by weeds, or I could potentially lose a villager without even knowing. The improved customisation options hooked me in for longer than usual, but I’m ashamed to admit that I haven’t loaded the game up for over a week now, and the longer I leave it, the less likely it feels that I’ll ever return. I know that my island will already be covered in weeds and the thought of collecting all of those apples, cherries, oranges and pears fills me with existential dread (there’s also the small fact that my real-world garden requires lots of care and attention right now, as my wife keeps reminding me).

It’s not that I’ve been hit with fatigue as much as fear; because I’ve had other games that have needed my attention (Lonely Mountains: Downhill, I’m looking at you), my limited game time in each day has been taken up elsewhere, and that means New Horizons has been pushed to the bottom of the pile. As I did with New Leaf, I’m sure I’ll pluck up the courage to load up New Horizons again soon, but my overriding fear is that I won’t have the stomach to fully commit myself to the cause every single day – and that’s really what’s needed for a game of this nature.

Perhaps the key issue here isn’t with the game – which offers a staggering amount of gameplay potential and content – but with me? Between running a network of sites, producing content, reviewing games and juggling the commitments of a house and family, my own ‘real-life’ game of Animal Crossing has robbed me of the time needed to truly enjoy Animal Crossing: The Video Game?

Gavin Lane, Features Editor

For me, Wild World was my first Animal Crossing and will likely always be my ‘favourite’. After devouring that game on DS, the idea of being tied down to a television seemed absurd. From my point of view, Animal Crossing was something you squeezed into your daily routine whenever and wherever you could, no matter what. You could be sitting on the bus or the train or the toilet and it wasn’t a problem – on a portable system you always had time pop into your village, check turnip prices and make sure your favourite residents weren’t packing their bags.

So, this is the first Animal Crossing I’ve played on big TV screen and every time I fire it up I’m still amazed at just how pretty it looks. PS5 and Xbox Series X won’t be losing any sleep, of course, but the lighting and attention to detail in New Horizons make it a pleasure to throw on the telly. Perhaps that’s a factor in why I’ve played it every day since launch – even at the expense of other games I’m itching to spend some real time with (hello Streets of Rage 4).

Much has been made of the fortuitous timing of New Horizons’ release, and it’s been a great help personally getting through this lockdown. Where pre-COVID-19 I might have gone for stroll of an evening to get some much-needed fresh air away from a computer monitor, I’ve found myself wandering around my island, idly catching bugs, fishing or arranging flower beds as a way to decompress at the end of the day.

In fact, it’s telling that I haven’t made much progress at all in terms of my house extensions or things like that. I managed to complete my fossil collection a couple of evenings ago, but that’s about it. I’ve been using the game more as a relaxation tool, and it’s in that capacity that I’m returning every night.

I can’t say whether I’ll still be playing when life returns to ‘normal’ again, but I’ve got more than my money’s worth from New Horizons in these past two months and I’m happy to potter about watering plants and making Star Trek uniforms. Lovely!

Alex Olney, Senior Video Editor

© Nintendo Life

Just like Gav upstairs my first dive into the wild world of Animal Crossing was on Animal Crossing: Wild World, although unlike his Gavvishness it’s not my favourite. Looking back it was great for the time, but it’s aged poorly and has naff all to do in comparison to the modern counterparts. Then it was New Leaf, and now it’s that other one that I’m supposed to be talking about.

Even though not all of the 150 hours I put in were entirely voluntary – having to play video games for a career is hard – I can honestly say that through the frustrations of trying to find certain fish and having to wait days for things to happen, I’ve enjoyed every one of those hours spent.

Suffice it to say though, the lustre is starting to lose its sparkle somewhat. What was a daily ritual every morning and most evenings has now become a dip-in-dip-out sort of affair. My island’s in largely good shape, I have some of the residents I want, but the drive to go back and get into the game isn’t as strong as it once was.

When I do pick it up though, I quite quickly tumble into a spiral of lost hours, even though when I booted it up I was certain I didn’t have much to do. Hell, last night after what was rather a frantic and exhausting day of more video games for money, I had so little energy for anything I just booted up the game and wandered around for a bit with no real purpose. And you know what? It was wonderful.

New Horizons may not have the vice-like grip it had on me before, and I may not be playing it as much as I once had, but I think it’s a game I’ll always be dipping in and out of, especially if these updates keep coming.

I hope this blog finds you, your family and friends and your colleagues all healthy and safe during these changing times. I’m excited to share some of the updates we are incorporating within Windows 10 to make it easier to see and use for people with low or no vision.  These improvements, coming to Windows 10 users in the May 2020 Update, represent the next steps in our journey to empower every person on the planet to achieve more.

Thank you for all the feedback to date. Please keep it coming! Our users, especially Windows Insiders, are continuing to shape Windows 10 accessibility by helping us understand what improvements matter most. Information about how best to reach us is included at the end of this post.

Making Windows 10 easier to see and use

Text cursor improvements

We recently made text and pointers easier to personalize in Windows 10. Millions of people prefer larger text, and larger and brighter pointers to make Windows 10 easier to see and use. Perhaps not surprisingly, many low vision users asked us to continue to make it easier to work with text cursors. More specifically, they told us that text cursors are difficult to find when they are too thin and don’t contrast enough with text. Now users can easily make cursors wider and add a text cursor indicator; we’ve even included the ability to choose custom colors, making it even easier to find their text cursor on the screen.

Figure 1 – New Ease of Access Settings make text cursors easier to see and use.

We made some related changes in Magnifier. We recently added a Magnifier option to follow the mouse pointer, so that users did not have to scan their display to find the pointer. Instead, they could rely on the mouse pointer to be in an expected location in the center of their display. Again, not surprisingly, users asked us to make it easier to work with text cursors with Magnifier. Now Magnifier will follow the text cursor in the center of the display by default, so that users can find the text cursor in an expected location.

Magnifier reading reduces strain and fatigue

We’ve heard from many Magnifier users that it can be difficult to read and work with text throughout their day; they often strain to read text and end their day with severe fatigue or headaches. We added a new feature that makes it easy to have text read aloud to reduce the strain of reading. Magnifier now includes Play, Pause, Next sentence, Previous sentence and Read from here controls to make it easy to read text in popular browsers including Edge, Chrome and Firefox and other Windows applications like Microsoft Word, Outlook and PowerPoint. Magnifier will highlight each word that is spoken and will scroll content into view if it is not currently displayed.

Finally, we’ve improved the Magnifier user interface. Magnifier will now respond to larger text and Dark mode settings, so that it is easier to see and use.

Figure 2 – New Magnifier features include reading and support for larger text and dark mode.

Making Windows easier to use without sight, with a screen reader

We’ve made a few sets of improvements to Narrator, which is the free, built in screen reader in Windows 10.

Narrator easier to use

We made reading more natural by reducing unnatural pauses and processing complete sentences to aid with proper pronunciation. We also redesigned Narrator sounds to enable Narrator users to be more efficient. We added sounds for the most common actions while reducing the total number of sounds to make them easier to learn and use. For example, rather than having to hear “scan on” or “scan off” when switching between scan mode and other modes, now you can choose to hear tones instead of the words. By default, both the words and sounds are turned on, to learn the new sounds. You can then change the level of detail that Narrator speaks about text and controls to hear fewer words for common actions. Finally, we improved support for announcing capital words and letters, e.g., when proofing a message or document. Narrator will handle initial, mixed and all cap words.

Smarter browsing

We made multiple improvements to the Narrator browsing experience.

First, Narrator will automatically start reading web pages from the top of the page. Automatic reading is intended to improve efficiency and give users confidence that the intended page loaded. Second, we implemented a page summary on demand. In addition to better understanding the complexity of a page, e.g., the number of landmarks, links and headings, you can also get a list of the most popular links generated by Bing to make it easier to get to where you want to go. Press Narrator key + S to hear the landmarks, links and headings on the page and press Narrator key + S twice quickly to bring up the page summary that includes a list of popular links. Third, we addressed one of the top accessibility issues with poorly crafted web pages. We made it possible to disambiguate links with titles like “Click here” without having to press the link to see where it goes. Press Narrator key + Ctrl + D, the same command to generate an image description, to hear the title of the linked page before pressing the link.

Finally, we made multiple improvements to Narrator with popular browsers. In addition to Microsoft Edge and Chrome, Narrator now supports Firefox. We improved several Narrator experiences, including faster “Find” and more reliable table reading. We also added support for rich text in Chrome and Firefox, which makes it easier to consume content on sites like Wikipedia.

More efficient Outlook mail

In addition to general Narrator and Narrator browsing improvements, we improved Narrator’s Outlook experience. We optimized Narrator for the Outlook inbox to make it faster and easier to triage mail; we read a smaller set of data in the expected order. We also improved the message reading experience. Like when web pages load, we now automatically start reading content when you open a message. We also automatically enable scan mode and recognize and ignore layout tables to make it easier to navigate and read message content. We improved Outlook responsiveness, e.g., reducing lags when arrowing through messages in the inbox or through text in a message.

Please keep the feedback coming

While we are excited to share our progress, we recognize that we have more work to do to create delightful experiences for people with disabilities. Thank you to the many people who have provided feedback — both positive and constructive — to help make Windows great. If you’re interested in providing help or suggestions, we welcome your feedback via the Windows Insider Program. All the features referenced in this blog were shared with and shaped by Windows Insiders. Whether you’re using an Insider build, or a generally available version of Windows, Windows 10 makes it easy to share your thoughts and suggestions — just press Windows logo key + F to launch the Feedback hub and share what’s top of mind.

Finally, if you are a customer with a disability and need technical assistance, the Disability Answer Desk is there to assist via phone (800-936-5900) and chat. In addition, we also have an ASL option available for our customers who are deaf or hard of hearing in the U.S. (+1 503-427-1234). Please contact us, we are always happy to help.

At the 2005 Conference on Neural Information Processing Systems, researcher Hanna Wallach found herself in a unique position—sharing a hotel room with another woman. Actually, three other women to be exact. In the previous years she had attended, that had never been an option because she didn’t really know any other women in machine learning. The group was amazed that there were four of them, among a handful of other women, in attendance. In that moment, it became clear what needed to be done. The next year, Wallach and two other women in the group, Jennifer Wortman Vaughan and Lisa Wainer, founded the Women in Machine Learning (WiML) Workshop. The one-day technical event, which is celebrating its 15th year, provides a forum for women to present their work and seek out professional advice and mentorship opportunities. Additionally, the workshop aims to elevate the contributions of female ML researchers and encourage other women to enter the field. In its first year, the workshop brought together 100 attendees; today, it draws around a thousand.

In creating WiML, the women had tapped into something greater than connecting female ML researchers; they asked whether their machine learning community was behaving fairly in its inclusion and support of women. Wallach and Wortman Vaughan are now colleagues at Microsoft Research, and they’re channeling the same awareness and critical eye to the larger AI picture: Are the systems we’re developing and deploying behaving fairly, and are we properly supporting the people building and using them?

Senior Principal Researchers Jennifer Wortman Vaughan (left) and Hanna Wallach (right), co-founders of the Women in Machine Learning Workshop, bring a people-first approach to their work in responsible AI. The two have co-authored upward of 10 papers together on the topic, and they each co-chair an AI, Ethics, and Effects in Engineering and Research (Aether) working group at Microsoft.

Wallach and Wortman Vaughan each co-chair an AI, Ethics, and Effects in Engineering and Research (Aether) working group—Wallach’s group is focused on fairness, Wortman Vaughan’s on interpretability. In those roles, they help inform Microsoft’s approach to responsible AI, which includes helping developers adopt responsible AI practices with services like Azure Machine Learning. Wallach and Wortman Vaughan have co-authored upward of 10 papers together around the topic of responsible AI. Their two most recent publications in the space address the AI challenges of fairness and interpretability through the lens of one particular group of people involved in the life cycle of AI systems: those developing them.

“It’s common to think of machine learning as a fully automated process,” says Wortman Vaughan. “But people are involved behind the scenes at every step, making decisions about which data to use, what to optimize for, even which problems to solve in the first place, and each of these decisions has the potential to impact lives. How do we empower the people involved in creating machine learning systems to make the best choices?”

Their findings are presented in “Co-Designing Checklists to Understand Organizational Challenges and Opportunities around Fairness in AI” and “Interpreting Interpretability: Understanding Data Scientists’ Use of Interpretability Tools for Machine Learning.” The publications received ACM CHI Conference on Human Factors in Computing Systems (CHI 2020) best paper recognition and honorable mention, respectively.

A framework for thinking about and prioritizing fairness

When Wallach took the lead on the Aether Fairness working group, she found herself getting the same question from industry colleagues, researchers in academia, and people in the nonprofit sector: Why don’t you just build a software tool that can be integrated into systems to identify issues of unfairness? Press a button, make systems fair. Some people asked in jest; others more seriously. Given the subjective and sociotechnical nature of fairness, there couldn’t be a single tool to address every challenge, and she’d say as much. Underlying the question, though, was a very real truth: Practitioners needed help. During a two-hour car ride while on vacation, Wallach had an aha moment listening to a Hidden Brain podcast episode about checklists. What practitioners wanted was a framework to help them think about and prioritize fairness.

“I’m getting this question primarily from people who work in the technology industry; the main way they know how to ask for structure is to ask for software,” she recalls thinking of the requests for a one-size-fits-all fairness tool. “But what they actually want is a framework.”

Wallach, Wortman Vaughan, Postdoctoral Researcher Luke Stark, and PhD candidate Michael A. Madaio, an intern at the time of the work, set out to determine if a checklist could work in this space, what should be on it, and what kind of support teams wanted in adopting one. The result is a comprehensive and customizable checklist that accounts for the real-life workflows of practitioners, with guidelines and discussion points for six stages of AI development and deployment: envision, define, prototype, build, launch, and evolve.

During the first of two sets of workshops, researchers presented participants with an initial AI fairness checklist culled from existing lists, literature, and knowledge of fairness challenges faced by practitioners. Participants were asked to give item-level feedback using sticky notes and colored dots to indicate edits and difficulty level of accomplishing list items, respectively. The researchers used the input to revise the checklist.

Co-designing is key

AI ethics checklists and principles aren’t new, but in their research, Wallach, Wortman Vaughan, and their team found current guidelines are challenging to execute. Many are too broad, oversimplify complex issues with yes/no–style items, and—most importantly—often appear not to have included practitioners in their design. Which is why co-designing the checklist with people currently on the ground developing AI systems formed the basis of the group’s work.

The researchers conducted semi-structured interviews exploring practitioners’ current approaches to addressing fairness issues and their vision of the ideal checklist. Separately, Wallach, Wortman Vaughan, and others in the Aether Fairness working group had built out a starter checklist culled from existing lists and literature, as well as their own knowledge of fairness challenges faced by practitioners. The researchers presented this initial checklist during two sets of workshops, revising the list after each based on participant input regarding the specific items included. Additionally, the researchers gathered information on anticipated obstacles and best-case scenarios for incorporating such a checklist into workflows, using the feedback, along with that from the semi-structured interviews, to finalize the list. When all was said and done, 48 practitioners from 12 tech companies had contributed to the design of the checklist.

During the process, researchers found that fairness efforts were often led by passionate individuals who felt they were on their own to balance “doing the right thing” with production goals. Participants expressed hope that having an appropriate checklist could empower individuals, support a proactive approach to AI ethics, and help foster a top-down strategy for managing fairness concerns across their companies.

A conversation starter

While offering step-by-step guidance, the checklist is not about rote compliance, says Wallach, and intentionally omits thresholds, specific criteria, and other measures that might encourage teams to blindly check boxes without deeper engagement. Instead, the items in each stage of the checklist are designed to facilitate important conversations, providing an opportunity to express and explore concerns, evaluate systems, and adjust them accordingly at natural points in the workflow. The checklist is a “thought infrastructure”—as Wallach calls it—that can be customized to meet the specific and varying needs of different teams and circumstances.

During their co-design workshops, researchers used a series of storyboards based on participant feedback to further understand the challenges and opportunities involved in incorporating AI fairness checklists into workflows.

And just as the researchers don’t foresee a single tool solving all fairness challenges, they don’t view the checklist as a solo solution. The checklist is meant to be used alongside other methods and resources, they say, including software tools like Fairlearn, the current release of which is being demoed this week at the developer event Microsoft Build. Fairlearn is an open-source Python package that includes a dashboard and algorithms to support practitioners in assessing and mitigating unfairness in two specific scenarios: disparities in the allocation of opportunities, resources, and information offered by their AI systems and disparities in system performance. Before Fairlearn can help with such disparities, though, practitioners have to identify the groups of people they expect to be impacted by their specific system.

The hope is the checklist—with such guidance as “solicit input on system vision and potential fairness-related harms from diverse perspectives”—will aid practitioners in making such determinations and encourage other important conversations.

“We can’t tell you exactly who might be harmed by your particular system and in what way,” says Wallach. “But we definitely know that if you didn’t have a conversation about this as a team and really investigate this, you’re definitely doing it wrong.”

Tackling the challenges of interpreting interpretability

As with fairness, there are no easy answers—and just as many complex questions—when it comes to interpretability.

Wortman Vaughan recalls attending a panel discussion on AI and society in 2016 during which one of the panelists described a future in which AI systems were so advanced that they would remove uncertainty from decision-making. She was confounded and angered by what she perceived as a misleading and irresponsible statement. The uncertainty inherent in the world is baked into any AI systems we build, whether it’s explicit or not, she thought. The panelist’s comment weighed on her mind and was magnified further by current events at the time. The idea of “democratizing AI” was gaining steam, and models were forecasting a Hillary Rodham Clinton presidency, an output many were treating as a done deal. She wondered to the point of obsession, how well do people really understand the predictions coming out of AI systems? A dive into the literature on the ML community’s efforts to make machine learning interpretable was far from reassuring.

“I got really hung up on the fact that people were designing these methods without stopping to define exactly what they mean by interpretability or intelligibility, basically proposing solutions without first defining the problem they were trying to solve,” says Wortman Vaughan.

That definition rests largely on who’s doing the interpreting. To illustrate, Wallach provides the example of a machine learning model that determines loan eligibility: Details regarding the model’s mathematical equations would go a long way in helping an ML researcher understand how the model arrives at its decisions or if it has any bugs. Those same details mean little to nothing, though, to applicants whose goal is to understand why they were denied a loan and what changes they need to make to position themselves for approval.

In their work, Wallach and Wortman Vaughan have argued for a more expansive view of interpretability, one that recognizes that the concept “means different things to different people depending on who they are and what they’re trying to do,” says Wallach.

As ML models continue to be deployed in the financial sector and other critical domains like healthcare and the justice system—where they can significantly affect people’s livelihood and well-being—claiming ignorance of how an AI system works is not an option. While the ML community has responded to this increasing need for techniques that help show how AI systems function, there’s a severe lack of information on the effectiveness of these tools—and there’s a reason for that.

“User studies of interpretability are notoriously challenging to get right,” explains Wortman Vaughan. “Doing these studies is a research agenda of its own.”

Not only does designing such a study entail qualitative and quantitative methods, but it also requires an interdisciplinary mix of expertise in machine learning, including the mathematics underlying ML models, and human–computer interaction (HCI), as well as knowledge of both the academic literature and routine data science practices.

The enormity of the undertaking is reflected in the makeup of the team that came together for the “Interpreting Interpretability” paper. Wallach, Wortman Vaughan, and Senior Principal Researcher Rich Caruana have extensive ML experience; PhD student Harmanpreet Kaur, an intern at the time of the work, has a research focus in HCI; and Harsha Nori and Samuel Jenkins are data scientists who have practical experience building and using interpretability tools. Together, they investigated whether current tools for increasing the interpretability of models actually result in more understandable systems for the data scientists and developers using them.

Three visualization types for model evaluation are output by the popular and publicly available InterpretML implementation of GAMs (top) and the implementation of SHAP in the SHAP Python package (bottom), respectively. Left column: global explanations. Middle column: component (GAMs) or dependence plot (SHAP). Right column: local explanations.

Tools in practice

The study focuses on two popular and publicly available tools, each representative of one of two techniques dominating the space: the InterpretML implementation of GAMs, which uses a “glassbox model” approach, by which models are designed to be simple enough to understand, and the implementation of SHAP in the SHAP Python package, which uses a post-hoc explanation approach for complex models. Each tool outputs three visualization types for model evaluation.

Through pilot interviews with practitioners, the researchers identified six routine challenges that data scientists face in their day-to-day work. The researchers then set up an interview study in which they placed data scientists in context with data, a model, and one of the two tools, assigned randomly. They examined how well 11 practitioners were able to use the interpretability tool to uncover and address the routine challenges.

The researchers found participants lacked an overall understanding of the tools, particularly in reading and drawing conclusions from the visualizations, which contained importance scores and other values that weren’t explicitly explained, causing confusion. Despite this, the researchers observed, participants were inclined to trust the tools. Some came to rely on the visualizations to justify questionable outputs—the existence of the visualizations offering enough proof of the tools’ credibility—as opposed to using them to scrutinize model performance. The tools’ public availability and widespread use also contributed to participants’ confidence in the tools, with one participant pointing to its availability as an indication that it “must be doing something right.”

Following the interview study, the researchers surveyed nearly 200 practitioners, who were asked to participate in an adjusted version of the interview study task. The purpose was to scale up the findings and gain a sense of their overall perception and use of the tools. The survey largely supported participants’ difficulty in understanding the visualizations and their superficial use of them found in the interview study, but also revealed a path for future work around tutorials and interactive features to support practitioners in using the tools.

“Our next step is to explore ways of helping data scientists form the right mental models so that they can take advantage of the full potential of these tools,” says Wortman Vaughan.

The researchers conclude that as the interpretability landscape continues to evolve, studies of the extent to which interpretability tools are achieving their intended goals and practitioners’ use and perception of them will continue to be important in improving the tools themselves and supporting practitioners in productively using them.

Putting people first

Fairness and interpretability aren’t static, objective concepts. Because their definitions hinge on people and their unique circumstances, fairness and interpretability will always be changing. For Wallach and Wortman Vaughan, being responsible creators of AI begins and ends with people, with the who: Who is building the AI systems? Who do these systems take power from and give power to? Who is using these systems and why? In their fairness checklist and interpretability tools papers, they and their co-authors look specifically at those developing AI systems, determining that practitioners need to be involved in the development of the tools and resources designed to help them in their work.

By putting people first, Wallach and Wortman Vaughan contribute to a support network that includes resources and also reinforcements for using those resources, whether that be in the form of a community of likeminded individuals like in WiML, a comprehensive checklist for sparking dialogue that will hopefully result in more trustworthy systems, or feedback from teams on the ground to help ensure tools deliver on their promise of helping to make responsible AI achievable.

The latest Nintendo Download update for Europe has finally arrived (Nintendo dropped the update one day later than usual), and it’s bringing new games galore to the eShop in your region. As always, be sure to drop a vote in our poll and comment down below with your potential picks for the week. Enjoy!

Switch Retail eShop – New Releases

Journey to the Savage Planet (505 Games, Thu 21st May, £24.99 / €29.99) WELCOME TO THE PIONEER PROGRAM! As the newest recruit of Kindred Aerospace – The 4th best interstellar space exploration company – Your job is to determine if the planet ARY-26 planet is fit for humans. You may be short on equipment and experience, but good luck!

Monstrum (Soedesco, Today, Today, £24.99 / €29.99) Stranded on a derelict cargo ship, you find yourself hunted by one of Monstrum’s terrifying predators as you search for a way out. Death is death. Get killed in Monstrum and you’ll be starting all over again. As each of them has their own strategy, abilities and weaknesses, you have to use your wits to outsmart your pursuer in the ship’s procedurally generated environment. Can you survive Monstrum?

The Persistence (Firesprite, Thu 21st May, £24.99 / €29.99) The Persistence challenges you to survive aboard a doomed deep space colony starship in the year 2521. Stranded, malfunctioning and caught in the inexorable pull of a black hole, “The Persistence” is overrun with a crew mutated into horrific & murderous aberrations. It’s down to you, a clone of security officer Zimri Eder, to make your way deeper in the decks of The Persistence to repair the systems and prevent the ship from being torn apart. Gather resources, upgrade abilities and fabricate an arsenal of weapons in this brutal sci-fi horror roguelike. Read our The Persistence review.

The Wonderful 101: Remastered (PlatinumGames, Tue 19th May, £39.99 / €44.99) A team of heroes from around the world must UNITE to protect the earth from vicious alien invaders! This band of 100 Wonderful Ones works together using their fantastic abilities to create a variety of forms. Whether it be a giant fist or a sharp blade, they’ll use their wits and power to overcome the enemy’s pitfalls and perils! And the final member of this team of courageous heroes—is you. Read our The Wonderful 101: Remastered review.

Switch eShop – New Releases

Arrest of a stone Buddha (CIRCLE Ent., Thu 21st May, £12.14 / €13.49) As a professional killer you’ll live through November 1976; when taking on enemies you’re fast and lucky enough so you don’t need to plan your actions too far. You kill with one shot and you never miss. When you run out of ammo you just throw a gun away. You can disarm any enemy and use any weapon; when it’s time to get out you just use a vehicle nearby.

Cannibal Cuisine (Rocket Vulture, 2020, £8.76 / €9.74) The god Hoochooboo is hungry… for you. But as all good cannibal cooks know, serving is better than being served. Tour the island Ready, steady, cook your way through 20+ levels in a full campaign across the island. Visit the jungle, temples, beaches and a lavalicious volcano!

Chess (Sabec, Thu 21st May, £8.09 / €8.99) Chess is the most famous strategic board game of all time. Easy to learn but difficult to master, play against the smart Ai (artificial intelligent) player and with four levels of difficulty to choose from anyone from a novice to a master can play. Chess is also playable against another player, either way winning won’t come easy.

Concept Destruction (Ratalaika Games, Today, £3.99 / €3.99) Concept Destruction is all about driving miniature cars made of cardboard, and crashing them into each other to earn points by destroying them! Pick from several different modes that suits your playstyle. Pick championship mode if you want to fight your way through mass production or choose survival mode to see how long you can survive a wave of deadly cardboard automobiles!

Fluxteria (Playstige Interactive, Thu 21st May, £6.29 / €6.99) Welcome to Fluxteria, non-stop arcade-space-shooting action in full 3D. Four modes of play: Story Mode, Survival Mode, Timed-Attack Mode and Obstacle Race Mode. 14 different types of Enemies. 20 different levels in beautifully stylized science-fiction setting. Crystal-Clear shooting and blasting sound effects. Hearth-thumbing music.

Golf With Your Friends (Team17, Tue 19th May, £14.99 / €19.99) Nothing is out of bounds as you take on 9 courses filled with fast paced, exciting, simultaneous mini golf for up to 12 players! Key Features: 12 Player Multiplayer! Make sure your skills are up to scratch as you tee off against 11 other golfers in online multiplayer.

Jigsaw Masterpieces (BottleCube, Thu 21st May, £0.89 / €0.99) Many convenient functions for completing the puzzle, such as the function to separate edge and inner pieces and the auto save function for playing lighthearted at any time.

Knight Swap 2 (QUByte Interactive, Tue 19th May, £0.89 / €0.99) Knight Swap is a minimalist puzzle game based on the old chess problem of swapping knights on a board. The goal in each level is to move the opposing knights to where the others are placed (and vice versa) using the features of the board.

Lost Artifacts: Time Machine (8Floor Games, Thu 21st May, £7.89 / €8.79) “A mysterious villain is using a time machine to destroy everything in his path. He wants to return the Priest, the King of Atlantis and the Ancient Emperor so he can use them to create a new world order! Claire and her friends must find out who the villain is and stop him and all his minions. Hurry, let’s go on a journey! Go on a journey across a world full of extinct creatures and fantastic technologies in the exciting casual strategy.

Luxar (KOALABS, Thu 21st May, £7.50 / €7.90) In LUXAR, you control an alien luminescent organic entity that has just been born and must pass its first survival trials to show its adult peers that it deserves to grow among them. It will have to cross 16 dark and hostile sectors with labyrinth-shaped alien architectures populated by toxic substances, obstacles, patrol drones and mechanisms of all kinds. You are its only chance to make it happen.

Monster Prom: XXL (Those Awesome Guys, Thu 21st May, £13.99 / €15.99) There’s only 3 weeks left until prom and you haven’t got a date yet. . and you’re a monster. But that’s okay ‘cause at your high-school everyone else is too! These are the stepping stones into the world of Monster Prom: XXL – the bundled up, content filled edition of Monster Prom, a one-to-four player competitive dating sim that will have you thinking: I never knew I was into that!

Myths of Orion: Light from the North (Ocean Media, Thu 21st May, £8.99 / €9.99) Long time ago, in a Realm of Orion, a powerful, but greedy wizard collected knowledge from all around the world. And thus came to write the three Books: books of Knowledge, Law and Magic. The books gave him ultimate powers. But, he used them not to prosper, but for chaos!

Pushy and Pully in Blockland (Resistance Studio, Thu 21st May, £9.99 / €9.99) Pushy and Pully in Blockland Push is a cooperative arcade game with a retro feeling. Help Pushy and Pully get back their spaceship on their trip across Blockland. Push blocks to defeat the monsters, join blocks of the same colour to get power ups and defeat the bosses to recover the parts of your broken spaceship.

Red Wings: Aces of the Sky (All In! Games, Thu 21st May, £17.99 / €19.99) Red Wings: Aces of the Sky is a dynamic action game that will show you amazing battles straight from the pages of history in a brand new arcade gameplay style. Use your unique skills to take down your opponents, develop your deadly pilot aim, and become the true ace of the sky. Witness the birth, rise, and fall of the Red Baron, Manfred von Richthofen—the ultimate hero of the first World War, credited with 80 victories.

Skelly Selest & Straimium Immortaly Double Pack (Digerati, Thu 21st May, £5.39 / €5.99) This bundle contains two wonderfully outlandish indie treasures from retro-inspired UK developer, Caiysware. Skelly Selest: a score-attack slash dash run ‘n’ gun-athon that places players in the skeletal shoes of a Heavenly Keeper. Only the Selestial Order can defeat the sinful horrors that lurk in the depths of Hell.

Steel Rain (PolarityFlow, Thu 21st May, £12.50 / €13.99) Prepare for the ultimate shoot-’em-up / strategy ReMix with RPG elements! Steel Rain is a truly novel mix of strategy, RPG, and frantic action, with a modern, high velocity shmup core at its center. You get to control 2 wings you can tactically arrange in different formations and firemodes, as you fight your way through an impressive variety of overwhelmingly huge enemy fleets. Liberate planets, build colonies, research new technologies, and learn new active and passive skills.

What The Golf? (TribandProductions, Thu 21st May, £13.49 / €14.99) A silly physics-based golf parody where every golf course is a new surprising type of golf, some brilliant or hilarious, others so absurd they will make you go: WHAT THE GOLF? Golf a house, a horse or car! Read our What The Golf? review.

Zenge (Hamster On Coke, Thu 14th May, £1.25 / €1.39) Zenge is a peculiar puzzle game, telling the story of Eon – a lonely journeyman who’s stuck between the worlds and time. Game is intended to be a relaxing experience, thus there are no points, stars, tutorials, move counters, in game shops or any other distractors. Just pure, immersive journey with Eon, told through gorgeous art and music.

Switch eShop – Demos

Switch eShop – Pre-Orders

Switch eShop – Special Offers

DLC / Add-On Content

Nintendo Switch games with new DLC this week:

So that’s your lot for this week’s Nintendo Download. Go on, be a sport and drop a vote in the poll above, and comment below with your hot picks!