Dear Museum Trustee

Dear Museum Trustee who has at least some belief in technology:

As you think about your next gift to the Museum of Fine Arts _______, can I make a suggestion? Would you believe me if I suggested that you could transform your one-time gift into a significant annual gift instead? I’m not suggesting that instead of giving $250,000 you give $25,000 for ten years — I’m saying that if you direct your gift in a certain way and only give it to a receptive director who wants to carry out an operating direction that is unlikely to happen without your gift as a provocation, you could write a check for $250,000 and within a couple of years, the museum’s annual operating budget would be $200,000 stronger and it would stay that way for as long as the administration wanted. I’m going to suggest that you to use your gift as a lever to change the institution and, by doing so, you will in effect being giving an annuity (even though your gift will be one-time). Yes, I am sure that the Museum would still love to have your collection or your named gallery, but if you want to give $250,000 that really matters, the single most effective way to support the museum is to sponsor a one-time project of data migration.

Wow, you say, are you serious? Could I possibly be suggesting that you direct such a significant gift to something so incredibly boring and behind the scenes and “in the weeds?” Visible activities of the museum — a new wing, a blockbuster exhibition –usually are the causes that attract major gifts because they raise the visibility of the institution (not to mention the donor). Philanthropic capital can — and does — build for the museum’s future by strengthening collections (and the institution’s reputation) and by replenishing capital stock (such as the building). But opportunity for significant philanthropic support to affect the operating budget of an institution are rare, since the operating budget is driven by the day-to-day work of the institution — day-to-day activities that erratically brings revenue in but has to consistently supports the costs of the staff and operations that carry out the daily work. On unusual occasions, capital gifts to endowment or to endowing the costs of certain programs or staff roles can support the operating budget by providing dedicated ongoing funds from that endowment, but besides funds for endowment, most big gifts don’t affect the never-ending struggle to balance the operating budget. Annual giving helps the operating budget (and usually is built in as a somewhat predictable component of revenue) but even a big gift to operating — while helpful today — is gone tomorrow.

Before explaining how and why I make this strange suggestion, I need to provide some background on what has happened and what is happening in museum technology to explain why I think funding data migration is the single best way for you to give today in a way that will financially benefit the museum for years to come.

Locally managed system #1: Collections Management System (CMS):

The core software system that all collecting institutions should have is known as a Collections Management System — this is the database where curators, registrars, art handlers, conservators, general counsels, insurance specialists and others enter and make use of data about objects in the collection. In the CMS, data are stored that track the most important information about the objects — when were they accessioned? When have they been loaned? Where are they physically located at this moment? Museums need to not know these essential facts about the objects in their care, and even if an institution doesn’t have a full CMS, it tracks these data somewhere (in a spreadsheet, in a database like Filemaker Pro or Access, or even in a filing cabinet). Tracking this ‘”inventory” is complicated work, your museum friends will remind you, and is not quite as simple as monitoring how many cases of Macaroni and Cheese are in the storage or on the shelf of your grocery story; museums care for unique objects with all sorts of history that matters (including the names of who donated them, you’ll be glad to know). Even libraries who (outside of their special collections) care for physical objects that they check out and check back in can rely on shared cataloging records to create most of their inventory. But the core of a museum’s work is to collect, preserve, research, and display unique objects — our institution’s version of a Dorothea Lange photograph or a Golzius print or a Japanese tea cup is different than yours. Setting up and managing this system to track the museum’s collections has long been a fundamental part of stewardship.

Locally Managed System #2: Digital Asset Management System (DAMs)

In the ’80s and ’90s a museum that invested in having a collections management system was doing what it needed to do in terms of technology investments to support its program. But in the early 2000s as museums began to document their collections with digital images, a new need emerged — for a system that would help them to manage the digital documentation of their collections. At first, keeping track of digital images wasn’t too big a deal. But museums care about quality — that is the business that they are in and the driving passion of the people who work there. Scans of old negatives soon led to passionate debates about the quality of the scan, and the priorities of those who cared most about the work of photography — the photographers — aligned with those who cared most about how the works in the collection were represented (curators and most other staff). Together they recognized that a scan did not magically recreate reality. In fact, it didn’t even re-create the quality of a print that had been approved for publication in the printed world. Pretty quickly museums found themselves swimming in raw scans, color corrected scans, cropped versions, different sized files (for until that last 5 years, it cost a lot to store and re-use the largest sized file), versions for print, for the web. These various files were stored on local servers, on hard drives, in conservation. They were stored off-line to tape or DVDs. Then they were re-scanned or re-shot as technology changed, the work was moved, or when it underwent conservation treatment. As they were generated, these files had names like “07435.TIF” that said nothing about the content of the file or the role that it played in a digital workflow. Digital assets proliferated.

It was a problem, but not a big one because the leadership of the museum didn’t really want to do much with these digital files anyway. Museum curators and directors and marketing VPs had a fair amount of anxiety about whether putting images, especially larger images, on the website would mean. Would people be satiated by the image and no longer come see the work or the exhibition? The creation of these files was primarily aimed toward supporting print publications (as they always had). From 2000 until the end of the first decade, the policy of putting images on the web was a source of ambivalence if not of fear to most museums. Even if bigger and better versions of the images were being created and kept somewhere within the museum, creating or using small and imperfect versions of the images for the tentative move to the web sufficed. Until they didn’t. As the years passed, museums moved from believing that showing the collection on the web was detrimental and dilutive to being essential. Museums now look to the web to announce their existence to the world. “Now where did we put those final, curator-approved versions of the image files?” “I think that Jack had them on his desktop.” “Oh, that’s right. But didn’t Jack take a new job last March?” “Hmmmn. I bet they’re here somewhere.”

Time for a quiz, reader, and please be honest: how are you doing managing your digital images at home? The ones from your SLR? And your daughter’s point and shoot? And your phone? And those ones that your brother sent you from Halloween? What file naming conventions do you use? Maybe you back them up on an external hard drive. And then you forget what you put on that drive so you back the whole thing up again, and tell yourself that when you have some time you’re going to weed the duplicates. And then your laptop died — what was on that again? Now imagine that managing those digital images was core to your work — to having your museum discovered and appreciated and supported in its efforts to attract foot-traffic, grants, and the community. Then, add to your dawning realization of the need for Digital Asset Management the recognition that the other digital files that you (or someone you contract with) owns are also part of what you need to manage before systems change and you lose them forever — audio guides, websites about particular exhibitions, images of donors at opening parties, videos or 3D capture of installations or the history of the museum. Suddenly, file names like “approved final version REALLY FINAL AS OF MAY 4 of Degas drawing accession number 20144216 with color corrections done by Yale UP but concerns over shading expressed by Conservation large derivative but not the largest file.JPG” probably won’t work much longer.

Over the past few years, museums invested in digital photography studios and the workflow around that process; leading museum technologists presented at conferences about how when a curator wakes up in the middle of the night and realizes that a drawing that she is studying was drawn in 1841 rather than 1842, she changes it on an app on her phone and that data is updated in both the collections management systems and, via a customized exchange between the Collections Management System and the Digital Asset Management System, it is updated in both and pushed to an XML repository that supports the museum website and when the museum website re-harvests the data again in a few hours, the revised attribution data will be updated on the public website. And when a publisher wants to use the raw files because for print they will use the CYMK color range rather than the RGB spectrum which was used to generate the version that’s on the website, the raw version of the image is “checked out” and sent. For most museums, this costly and difficult-to-implement and maintain infrastructure is a dream.

Unfortunately, this dream as target for a sophisticated digital infrastructure effort is being carried out in old fashioned ways, defined by the local customs. DAMs implementations at museums may have a thousand problems — too costly, too much customization, too much blind hope that these systems which are primarily aimed at repository services will solve the myriad workflow issues associated with building digital collections — but the biggest problem is that they are being implemented locally. Some of the software systems being deployed are expensive and some are free (widely joked about as “free as in free puppies not as in free beer.”) But whatever software they are choosing to buy or build, if they are maintaining it locally, it is the wrong answer. By doing so, the institution is committing itself to the staffing that maintains the software, customizes the data exchanges between the CMS and the DAMs and any other system every time any other element of the local software ecosystem changes, modifies the data in other ways, develops network security around the software. It’s a crazy amount of ongoing technology support for services that need not be tailor-made.

It is inevitable that the key digital infrastructure upon which a museum depends for managing its collections and the documentation of those collections will move to the cloud. While some of you know what “the cloud” means (and others among you would rather be reading about Constable’s clouds), it is fairly simple to explain what “the cloud’ does. By using software that is based outside of the institution, on infrastructure that is managed on the network, provided and cared for by external agents rather than installed locally, your museum will — — sooner or later — be part of the networked world that has chosen to pay annual fees for technology services rather than the costs of acquiring, modifying, and maintaining software locally. Ten years ago, most businesses were reluctant to entrust any data, let alone sensitive data to be managed off site in this way. Paychecks were processed and printed on site; now, the odds are that your paychecks are managed remotely on The old “buy versus build” debate has mostly become a “buy cloud services or build cloud services” debate for most companies, and for many functions, even sensitive ones, the answer has been to buy cloud services. Or really, to rent them. But museums have not moved in this direction for their digital databases. But they will. They definitely will.

So why don’t they do so now, as they embark on the DAMs investment? A number of perfectly understandable historical and current pressures lead museums to think and act locally. Your institution, like most others, was born and lived for the most part in isolation. The museum implemented its collection management system in ways that had little to do with how other museums were managing their collections and everything to do with the preferences and previous ways that the museum processed objects. In this way, even when museums selected the same software vendor for managing collections, they would implement the system in different ways. What data mattered to the curator in a particular department in 1994? Were measurements entered as one field or were individual measurements parsed? Was the national origins of an artist entered or was the primary factor the location where the work was created? As they embark on implements a solution for managing digital assets, they are doing what they have always done — solving a mission-critical problem by constructing a local system within the control of the staff to manage. Which fields from the CMS matter for inclusion in the DAMs? Were the connections between the systems one-way or two-way streets? In other words, could someone who preferred the data entry screen in the DAMs use that? Would technical data drawn from the camera be copied into the CMS from the DAMs?

Migration or “Now the whole idea gets really tiring.”

After a museum chooses a system and integrates it with other systems, the job of migrating the data — literally getting it out of where it currently lives and into the right place in the new system is surprisingly hard. Computers don’t actually talk to computers. Part of the reason is that all data loading is difficult and the more complex the data are, the harder the loading. The process of extraction, transformation, and loading (ETL) requires understanding the source data and what needs to happen to it before in can be moved into a new database. This includes understanding the constraints of each field (is it numeric or alphabet characters? Are there limits to the number of characters? Does it require a controlled value — that is, a value that is entered from a list of pre-approved entries?)

And that’s when things are “simple” (this is actually a technical term — and one that’s opposed to “complex”). A complex object is defined by having relationships with other objects. So, a spoon in a Victorian tea set is an object — an accessioned work — but it is also related to the other objects in that tea set. Or a photo of Matisse in his studio is a work in itself but it also relates to the artists or to the other works visible in that photograph. The flexibility of databases allow these relationship between works to be expressed as links between tables in the database. That’s all fine and good until it’s time to recreate those relationships in a new database (if one were to migrate out of the current CMS and into another or if one is trying to customize a link between a new DAMs and an existing CMS). The migration of data inevitably ends up as an iterative process of trial and error as these relationships are uncovered. There’s no magic migration tool. There are scripts that are written and re-written and tested and revised to translate data from one set of database tables to another.

Moreover, while the data might have been entered into a software program, they were entered by human beings — people with different preferences, habits, or training. Extracting data and preparing it to be mapped inevitably surfaces discrepancies in the data entry processes “Back in ’03 we used to use that field to store provenance data before we added a new field in ’08.” Ask any of your friends who work in technology or have been at an institution that took on migrating legacy data: if those data had been entered by hand, over time, there will be quirks — quirks that only show up when the data is laid out for all to see by the migration process. The migration process will be delayed as staff debate about whether the heterogeneous data should be cleaned up in the mapping process or cleaned up by hand (either before migrating or after). But, again, the process will create more work.

In this work, debates about deeply held passions will be surfaced around “lossiness.” Pragmatists will urge that staff settle for compromise — if a certain field is messy, just drop it. If complex objects are too hard to understand, leave them out. If data pertaining to a certain segment of the collection was approved by the previous curator in a department but not by the current one, so be it. But the pragmatists will have to convince some people who have chosen a career in cultural institutions because they value excellence over expediency. Don’t expect these conversations to be concluded easily — at least unless leadership invests time in understanding the issues and setting direction.

Why this matters?

As a trustee, your first responsibility is as a fiduciary of the institution. You may be more (or less) involved in setting or approving strategic directions. You are certainly mindful of your charge to hire, review, and when necessary to replace the CEO. But, in your first and clearest responsibility you are charged with overseeing the financial well-being of the institution. And as a fiduciary of a non-profit institution, you probably have been a little confused about what to do about technology.

On the one hand, you have heard (maybe since childhood!) that technology would save us all money. It might be a little confusing that that never seems to have worked out. Then, as you and your colleagues watched the investment in the technology budget grow and grow, you hoped that there would be some corresponding growth in revenues — and you may even have been encouraged to hope for this at some point when someone told you about how special your museum’s collection is and how digital access would lead to growth in licensing fees.

Then you stopped hearing that. You now hear that there’s a tremendous flood of digital content out there and that more and more of the images of artworks in the public domain are being given away free of charge. While someone writing a scholarly article will need an image of a particular work, to someone selling playing cards or an umbrella, the free Monet looks a whole lot more appealing than the one that has to be licensed for a fee. So, now you’re hearing that digital content is a mission-related good.

I would propose that we face up to the reality that, yes, it is really important that the museum document the collection and the life of the museum and doing so will be an ongoing responsibility. The reasons are both known and un-known. It isn’t just to be represented on the website or in the Google Cultural Institute or Artsy or the Digital Public Library of America or Artstor or any other aggregation.

You want to be in the ballgame of getting your content into the flow. Yes, you want scholars to study and teach from the works in your collection. Yes, you want to be in Google searches and Streetview and Yelp and Trip Advisor. It will continue to be important that people coming to your city or trying to figure out which city to visit can see all the wonders that your museum cares for. But the ways that images are just starting to be woven into our digital lives is still in the very early innings. To give just one example — the growing field of in-image and in-screen advertising (see for example the work of is algorithmically matching advertisers’ visual content with news (and pseudo-news) content. In a news story about sharks, you might find an ad for visiting Sea World. With 3 billion images being uploaded to Snapchat, Instagram, and the like every day, we will — inevitably — become a world of image “readers.” And the images of the unique monuments of our cultures and our achievements will and should have a central place in that world. But they won’t if we steward costly — and walled — gardens.

As a trustee, you cannot make this all generate a balanced operating statement. And it may well be that the museum administration — for all of their many strengths — are not going to be the ones to figure out a revenue-generating plan that has any link to reality. That leaves them — and you — one lever: cost.

It is always more fun to anticipate future revenues than to reduce current operating costs. That is why universities build new state-of-the-art weight-training rooms (in anticipating of luring recruits who will get the team on TV’s “Game of the Week.” And non-profits face the challenging resource allocation rubric of knowing that sometimes it is core to the institution’s mission to spend money even if future revenues don’t appear. Without this mission sensitivity no college in the world would teach Coptic. But the first step to not spending what you don’t have to spend is to know the difference between a mission-differentiating function and a commodity. If you take nothing else away from reading this piece, my core message would be this: managing the two key systems that track your collections and your digital documentation of those collections should not be considered a mission-differentiating function. It should not require local software, hardware, or database engineers on an ongoing basis. It should not require ongoing customization. It should not require network engineers and firewall experts. If companies all over the world are depending on software-as-a-service to manage mission critical data, laden with privacy concerns, your institutions should not be hiring and paying for ongoing technical staff to track digital files. There are many barriers to “getting there” — some are social, some are intertwined with various institutional histories, so are sensitivities, and others — understandably — are tied up with the anxiety associated with reducing staff costs (though, of course, these can happen over time via attrition and need not be thought of as layoffs). But, to return to the topic which seemed so strange at the beginning of this letter, the onerous and inevitably costly work of migrating systems is a very real barrier to what otherwise could be a significant cost reduction and program-enabling move to cloud-based digital infrastructure. Some enthusiasts for such a move might want to wish away these costs and ask “how hard can that be?” It will be hard and it will take time and effort of various knowledgeable staff. Others might respond that it’s too hard. That has often been one of a series of justifications for doing nothing. If you can overcome the “too easy” crowd (who might argue that your dollars should go elsewhere) and the “too hard” crowd who want to wait for another day to do this, even if it means buying or building more locally maintained software in the meanwhile, you could use the lever of your gift to lift the museum in highly impactful ways.

A few words of caution

Conditions can be better or worse for using this lever. The migration process that I have advocated for is not something that the museum will want to do again soon and one danger of the cloud (to mix metaphors) is its stickiness. Once the museum gets its CMS data and its DAMs data into a cloud-based solution, it may be difficult to get your data out. For-profit cloud providers don’t want to help you — did Box ever offer to bulk download your files so that you could move them to DropBox? Does Yahoo.mail offer to deliver your archived emails from the last decade into Gmail for you? There have been attempts at non-profit consortial cloud-based systems (see archivespace, collectionspace, hydra-in-a-box as hosted by Duraspace, Shared Shelf). Are they good enough? I don’t know, but I theory at least, it would better if the museum felt that it was choosing a mission-aligned partner where fees went to support ongoing development rather than corporate profits. Or maybe a group of museums will have to band together to create together or work with a trusted third party to create what they need. But that might not always be the best option, so working with commercial firms might be the most cost effective path; if so, agreeing upon “pre-nuptial” terms by which the Museum can get its content back in a useful form would be important.

The second potential problem is if you pay for this migration and there’s no effort to stick to a plan to reduce technology staffing accordingly (over-time). There will always be plenty for tech staff to do. If you want to do prompt this effort in the efforts of avoiding or reducing costs on the operating budget, you need to press for a plan and a commitment to this. Of course things may change. But if reducing the amount of money that the Museum spends on on-site technology staff is not part of the plan, you may still end up serving the Museum well (by getting the content into a cloud-based service that may enable you to play better in a networked world); but without a plan to focus on reducing costs, it would be naïve to think that it would happen. Not out of any bad motives — the fuzziness of institutional mission (Peter Drucker famously called a mission statement, “a hero sandwich of good intentions”) would justify the continued investment in any number of technology projects. But, remembering that it is unlikely that playing in the networking of digital content will directly contribute to revenue in any significant way, your Museum will only capture your gift as a good financial investment if the leadership of the Museum sees the goal as being to truly offload the technology effort of this aspect of digital stewardship to others who can — at scale — keep up with technological change in a way that would not be efficient for the Museum’s own staff to do.

Give them a reason to move forward

Museums conserve. They are conservative. This is a very good thing. I don’t mean that to belittle all of the ways that they have innovated and moved forward and led their communities. But, at their essence, their mission is to conserve. And protect. And their default ethos is to avoid compromising on quality whenever possible. This proposal respects those traditions and that mission. But, in one small corner, away from the objects and the building that contains and displays them, there is an important future dressed up as a cost center. You can set them free to move there, to get the cataloging information and images that will awaken curiosity and allow new audience’s passions to be nurtured, while sparing the museum the cost of figuring out (again and again) how to maintain the infrastructure that enables them to do so. Challenge the place, and in doing so, help to strengthen it.

Show your support

Clapping shows how much you appreciated jamesshulman’s story.