Get ebook downloads in full at ebookmeta.

com

Machine Learning Governance for Managers 1st

Edition Francesca Lazzeri Alexei Robsky

https://ebookmeta.com/product/machine-learning-governance-
for-managers-1st-edition-francesca-lazzeri-alexei-robsky/

OR CLICK BUTTON

DOWNLOAD NOW

Explore and download more ebook at https://ebookmeta.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Self Service Data Analytics and Governance for Managers

1st Edition Nathan E Myers Gregory Kogan

https://ebookmeta.com/product/self-service-data-analytics-and-
governance-for-managers-1st-edition-nathan-e-myers-gregory-kogan/

ebookmeta.com

Practical Machine Learning for Computer Vision: End-to-End

Machine Learning for Images 1st Edition Valliappa
Lakshmanan
https://ebookmeta.com/product/practical-machine-learning-for-computer-
vision-end-to-end-machine-learning-for-images-1st-edition-valliappa-
lakshmanan/
ebookmeta.com

Machine Learning for Cloud Management 1st Edition Kumar

https://ebookmeta.com/product/machine-learning-for-cloud-
management-1st-edition-kumar/

ebookmeta.com

The Psychology of PCOS: Building the Science and Breaking

the Silence 1st Edition Stacey L. Williams

https://ebookmeta.com/product/the-psychology-of-pcos-building-the-
science-and-breaking-the-silence-1st-edition-stacey-l-williams/

ebookmeta.com
Teche A History of Louisiana s Most Famous Bayou Shane K
Bernard

https://ebookmeta.com/product/teche-a-history-of-louisiana-s-most-
famous-bayou-shane-k-bernard/

ebookmeta.com

Navigating Disruption Media Relations in the Digital Age

1st Edition Bertrand Teo

https://ebookmeta.com/product/navigating-disruption-media-relations-
in-the-digital-age-1st-edition-bertrand-teo/

ebookmeta.com

Bar Hebraeus The Ecclesiastical Chronicle: An English

Translation David Wilmshurst

https://ebookmeta.com/product/bar-hebraeus-the-ecclesiastical-
chronicle-an-english-translation-david-wilmshurst/

ebookmeta.com

Gilles Deleuze and Film Criticism Philosophy Theory and

the Individual Film 2nd Edition Ivan A. Ross

https://ebookmeta.com/product/gilles-deleuze-and-film-criticism-
philosophy-theory-and-the-individual-film-2nd-edition-ivan-a-ross/

ebookmeta.com

Writing Papers in the Biological Sciences Seventh Edition

Victoria Mcmillan

https://ebookmeta.com/product/writing-papers-in-the-biological-
sciences-seventh-edition-victoria-mcmillan/

ebookmeta.com
Process Design Integration and Intensification Mahmoud El-
Halwagi (Editor)

https://ebookmeta.com/product/process-design-integration-and-
intensification-mahmoud-el-halwagi-editor/

ebookmeta.com
Francesca Lazzeri and Alexei Robsky

Machine Learning Governance for

Managers
Francesca Lazzeri
Microsoft Corporation, Newton, MA, USA

Alexei Robsky
Google, Sammamish, WA, USA

ISBN 978-3-031-31804-7 e-ISBN 978-3-031-31805-4

https://doi.org/10.1007/978-3-031-31805-4

© The Editor(s) (if applicable) and The Author(s), under exclusive

license to Springer Nature Switzerland AG 2024

This work is subject to copyright. All rights are solely and exclusively
licensed by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in
any other physical way, and transmission or information storage and
retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.

The use of general descriptive names, registered names, trademarks,

service marks, etc. in this publication does not imply, even in the
absence of a specific statement, that such names are exempt from the
relevant protective laws and regulations and therefore free for general
use.

The publisher, the authors, and the editors are safe to assume that the
advice and information in this book are believed to be true and accurate
at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, expressed or implied, with respect to the
material contained herein or for any errors or omissions that may have
been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer
Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham,
Switzerland
Introduction
In today’s digital age, machine learning has become a game-changer for
businesses looking to gain a competitive advantage. From predictive
analytics to chatbots and robotic process automation, machine learning
algorithms are being applied to various domains, including finance,
healthcare, e-commerce, and customer service. By automating routine
tasks and analyzing vast amounts of data, machine learning can help
businesses optimize their operations, increase efficiency, and provide
better customer experiences.
However, as machine learning becomes more prevalent, the need for
effective governance of these systems becomes more critical. The
consequences of poorly governed machine learning can be severe, from
biased decision-making to privacy violations. Without proper
governance, machine learning can also learn the wrong outcomes,
leading to unintended consequences that can undermine the trust and
reputation of your organization and hurt business success.
Most organizations have access to more data than ever before.
However, the sheer volume and complexity of data can be
overwhelming, and many organizations struggle to unlock insights from
the data. Even when insights are identified, many organizations
struggle to operationalize them and put them into action. This is where
machine learning can come in. By leveraging algorithms and
automation, machine learning can help organizations make sense of
their data, identify patterns and trends, and make more informed
decisions.
However, implementing machine learning is not without its
challenges. One of the biggest challenges is operationalizing machine
learning models and incorporating them into business operations and
decision-making processes. Many organizations struggle with this step
because it requires not only technical expertise but also a deep
understanding of the business context and goals and the ability to
communicate insights effectively to stakeholders. Furthermore, even if
insights from machine learning models are successfully
operationalized, organizations still need to leverage them effectively to
improve customer experience, optimize their services, automate their
operational processes, and realize bottom-line gains. This requires a
framework, which is described in this book, to leverage AI solutions,
which includes not only technical considerations but also business
strategy, organizational culture, and change management.
In particular, 87% of organizations struggle to maintain a
sustainable machine learning model lifecycle. This includes everything
from model selection to training, validation, and testing, as well as
ongoing monitoring and maintenance. Moreover, deploying machine
learning models can be a time-consuming process; for 64% of
organizations, it takes a month or longer to deploy a single model
(Algorithmia Survey Report 2021) and build consumable and scalable
AI applications on top. This can be attributed to a variety of factors,
including the need for extensive testing and validation, the complexity
of the model architecture, and the lack of a standardized deployment
process. Furthermore, once a model has been deployed, organizations
need to ensure that it is scalable and can be integrated into existing
systems and processes. To address these challenges, organizations need
to think more organically about the end-to-end data flow and
architecture that will support their data science solutions.
Organizations need to ensure their machine learning solutions are
aligned with their business goals, they have the right talent and
resources to support them, and they have a clear understanding of the
regulatory considerations of using AI.
As organizations increasingly rely on data and AI to drive their
business operations and outcomes, the role of data science or business
managers has become more critical. Managers in that field are
responsible for overseeing the entire machine learning process, from
identifying business requirements to model deployment and
management. However, managing the end-to-end machine learning
process can be challenging, particularly when it comes to ensuring that
machine learning solutions are scalable, sustainable, and aligned with
existing IT and privacy policies. Our framework provides a portfolio of
methodologies, technologies, and resources that will assist managers in
scaling their machine learning initiatives and becoming more data and
AI driven. This includes tools for data generation and acquisition,
model selection and development, and model deployment and
management. Our framework emphasizes the importance of testing all
models, creating the right documentation, and monitoring models and
their results and causal business impact.
Versioning models is essential for maintaining a historical record of
all changes made to the model, which helps to ensure reproducibility
and traceability. Creating the right documentation, including model
architecture and design, can help managers to communicate effectively
with stakeholders and ensure that models are understandable and
transparent. Lastly, monitoring models and their results can help to
identify potential issues and ensure that models are delivering the
expected business impact.
Effective AI/ML model governance is essential for organizations
that want to maximize the benefits of their AI investments.
Organizations that effectively implement all components of AI/ML
model governance can achieve a fine-grained level of control and
visibility into how models operate in production while unlocking
operational efficiencies that help them scale and achieve higher ROI
with their AI investments. By tracking, documenting, monitoring,
versioning, and controlling access to all models, these organizations can
closely control model inputs and understand all the variables that
might affect their results.
What is our framework about? By using this framework (Fig. 1),
managers will learn:
How to formulate business objectives and translate to measurable
outcomes?
How to establish performance metrics that are linkable to business
objectives?
How to leverage machine learning open-source frameworks and
toolkits to accelerate the model lifecycle?
How to design end-to-end machine learning solutions by making ML
technologies, programming languages, or frameworks compatible
and integrated into one architecture?
How to implement a machine learning model governance to control
access, implement policy, and track activity for models?
How to unify organizations’ machine learning vision? Successful AI
initiatives require organizational alignment across multiple decision-
makers and business functions.
Chapter 1—Understanding Business Goals
Business goals are an essential part of establishing priorities and
setting your company up for success over a set period. Taking the time
to set goals for your business and create individual objectives to help
you reach each goal can greatly increase your ability to achieve those
goals.
In this chapter, we explore the different ways to understand and act
upon business goals. One might say that each business knows their own
goals and has a solid strategy to achieve those goals. Setting goals
without a clear plan can result in wasted resources and ineffective
efforts. One common issue that companies face is the tendency to target
everything and nothing all at once. This can happen when there are too
many competing priorities or when teams try to tackle too many
objectives at once. This approach can lead to diluted efforts, which
ultimately hinder progress toward achieving the overall business goals.
It is not always easy to establish the right goals. Leaders may have
different opinions on what the priorities should be, or external factors
may change, requiring a shift in the goals. However, this is not to
discourage leaders from establishing goals but rather to provide a
different perspective on measuring what is right. It is crucial to
continuously evaluate the goals and adjust them as necessary to remain
aligned with the company’s overall strategy. Successful companies have
a clear understanding of their purpose and direction. Goals and
objectives play a crucial role in providing that clarity. Goals are general
statements of desired achievement, while objectives are the specific
steps or actions that a company takes to reach their goal. In other
words, goals establish the direction and vision of the company, while
objectives provide the necessary action plan to achieve those goals.
Both goals and objectives should be specific and measurable. It is
essential to establish goals and objectives that can be quantified and
tracked through data and data science solutions. By doing so, a
company can establish a clear baseline for their progress toward
achieving their goals and objectives, identify areas of improvement, and
adjust their strategy accordingly.
Table 1 summarizes different dimensions for which goals can be
established. The dimensions can range from profitability and growth to
customer service and employee satisfaction. For example, a company
may establish a goal to increase profitability. The objective to achieve
that goal may include increasing sales revenue or reducing costs. Goals
and objectives can be established at different levels of the organization.
For instance, a company may have a company-wide goal to increase
revenue, with department-level objectives to achieve that goal, such as
increasing product sales or improving marketing campaigns.

Fig. 1 ML governance framework

Table 1 Dimensional overview of goals and objectives

Dimension Goal Objective

Scope End result A mean to an end
Foundational Vision Data and facts
process
Magnitude Large Many metrics to support the
larger goal
Dimension Goal Objective
Evaluation Goals are usually inspirational Specific data and measurable
and intangible outcomes
Chapter 2—Measuring What Is Relevant
As discussed in the previous chapter, it is essential for a business to
understand its goals, tie them to measurable outcomes, and establish
the right monitoring, such as OKR, to evaluate Data Science and
machine learning efforts. In this chapter, we go deeper on performance
measurements and metrics and emphasize on understanding second-
level metrics. Measurement is the process of associating numbers with
physical quantities and phenomena. In the context of data science and
machine learning, measurement is fundamental to evaluating the
accuracy and usefulness of models and algorithms. Measuring the
performance of models and algorithms requires identifying the
appropriate metrics to use.
Metrics are quantitative measures used to evaluate performance.
Choosing the right metrics is critical to accurately assessing the
performance of data science and machine learning efforts. While many
organizations focus on primary metrics such as revenue, customer
acquisition, or cost savings, it is equally important to understand
second-level metrics. Second-level metrics are those that are directly
linked to primary metrics and provide insights into how well the
business is performing in a particular area.
For example, if a business’s primary metric is revenue, the second-
level metrics could be average order value, customer lifetime value, or
conversion rate. By understanding these second-level metrics,
businesses can identify the factors that contribute to the primary
metric and make data-driven decisions to improve performance.
The relevance of data is essential for any business or organization
that aims to leverage its data and make informed decisions. It can be
summarized in three dimensions, as discussed before:
1. Appropriateness of business concepts and their measurement
through data. Data scientists need to ensure that the data they use
aligns with the business concepts that the organization aims to
measure. In other words, they need to ensure that they are
measuring the right things and that the measures are relevant to
the business context. This requires continuous review and revision
of concepts and measures as problems change, and as analysis
reveals weaknesses in current measures and suggests alternate
 measures that might better capture information of use to
constituents.
2. Ability to link data. Data scientists need to ensure that they are
using data that can be linked to other data sources, both internal
and external. The ability to link data collected through various
instruments and to external data sources increases the breadth and
depth of data, and thereby, the ability of data scientists to use them
to address current issues. This means that data scientists need to
ensure that the data they use is structured in a way that enables
easy linkage with other data sources.

3. Data currency. This refers to whether data reflect current

conditions, such as seasonality, trends, accuracy, and speed of the
events under study. Data scientists need to ensure that the data
they use is up-to-date and accurately reflects the current conditions
that they are studying. This requires regular monitoring and
updating of data to ensure that it remains current and relevant.
Chapter 3—Searching for the Right Algorithms
Machine learning is a powerful tool that can help organizations
transform their business operations, improve decision-making, and
create new revenue streams. In this chapter, we will learn what
machine learning is and how being a machine learning organization
implies embedding machine learning teams to fully engage with the
business and adapting the operational support of the company. Most
importantly, we will learn the following four dimensions that
companies can leverage to become machine learning-driven:
Understanding Algorithms and the Business Questions that
Algorithms can answer. We discuss the importance of understanding
algorithms and how they can be used to solve specific business
questions. Machine learning algorithms are designed to learn from
data and make predictions or decisions based on that data. To
leverage the full potential of machine learning, companies need to
understand the various types of algorithms and their applications to
different business problems.
Defining Business Metrics and Business Impact. We focus on defining
business metrics and measuring the impact of machine learning on
business outcomes. Business metrics are key performance indicators
(KPIs) that can be used to measure the success of machine learning
initiatives. It is important for companies to define these metrics and
track them over time to understand the impact of machine learning
on their business.
Establishing Machine Learning Performance Metrics. We will delve
into establishing machine learning performance metrics. These
metrics are used to evaluate the performance of machine learning
models and ensure they are meeting the required accuracy,
reliability, and efficiency standards. Companies need to establish
clear performance metrics and continuously monitor and optimize
their machine learning models to ensure they are delivering the
expected results.
Architecting the End-to-End Machine Learning Solution. We discuss
the importance of architecting an end-to-end machine learning
solution. A successful machine learning implementation requires a
holistic approach that considers all stages of the machine learning
 lifecycle, including data acquisition, model development,
deployment, and monitoring. By considering the end-to-end process,
companies can ensure that their machine learning solution is
scalable, efficient, and can deliver the expected results.
Moreover, in this chapter, you will learn how to navigate the data
science and machine learning tool landscape. The data science and
machine learning tool landscape is constantly evolving and there are
several important aspects of tools to consider when building a machine
learning organization. One is the category. There are two important
categories of data science tools we highlight: open-source versus
proprietary code and no/low code versus code. Many of the most
popular data science tools are open source, meaning that they are free
to use and the underlying code is available for anyone to view and
modify. However, there are providers that use proprietary code. Even
providers that rely on proprietary software open source either some
parts of their solution or launch initiatives to attract data scientists that
would like to work on an open-source platform. Another important
distinction to consider is whether a tool requires coding skills or can be
used by individuals with no or low coding experience. No/low code
tools often provide a user-friendly interface that allows users to
conduct data analysis, experimentation, and machine learning without
needing to write complex code. This can be beneficial for individuals
with less technical expertise or those who need to quickly iterate on
experiments and projects. We discuss some data science tools that data
scientists use to conduct data analysis, experimentation, and machine
learning, and we will understand the main features of the tools, their
benefits, and the comparison of different data science tools.
Chapter 4—Operationalizing Your Machine
Learning Solution
Machine learning and data science are tools that offer companies the
possibility to transform their operations: from applications able to
predict and schedule equipment’s maintenance, to intelligent R&D
systems able to estimate the cost of new drug development, to HR AI-
powered tools able to enhance the hiring process and employee
retention strategy. However, to be able to leverage this opportunity,
companies must learn how to successfully build, train, test, and push
hundreds of machine learning models in production, and to move
models from development to their production environment in ways
that are robust, fast, and repeatable.
In this chapter, we explore some common challenges of machine
learning model deployment and we discuss the following points in
order to enable you to tackle some of those challenges:
Why successful model deployment is fundamental for data-driven
companies? Deploying a model successfully means ensuring that the
model produces accurate and reliable results in production.
Why companies struggle with model deployment, such as difficulties
in scaling the models, ensuring reliability, and dealing with technical
debt?
How to select the right tools to succeed with model deployment? The
right tools help companies streamline the deployment process,
minimize human error, and automate repetitive tasks. We examine
the features that companies should look for when selecting tools for
model deployment, such as support for different programming
languages, scalability, ease of use, and integration with other tools.
A few key aspects that companies need to solve to overcome the
barriers in terms of ML operationalization are:
1. Data Accessibility—a crucial factor for companies to succeed in
today’s data-driven world. A well-defined data and machine
learning strategy can help companies to define where business data
is stored, in what format, which quantity, and how to get access to
these in an automated environment in production. A clear data
Other documents randomly have
different content
more remarkable is that in his poetry he seems to feel the influence
of women mystically. I shall have to discuss this topic in another
place. It is enough here to say that, with very few exceptions, we
remain in doubt whether he is addressing a woman at all. There are
none of those spontaneous utterances by which a man involuntarily
expresses the outgoings of his heart to a beloved object, the throb of
irresistible emotion, the physical ache, the sense of wanting, the joys
and pains, the hopes and fears, which belong to genuine passion....
Michelangelo’s ‘donna’ might just as well be a man; and indeed, the
poems he addressed to men, though they have nothing sensual
about them, reveal a finer touch in the emotion of the writer.”
(Life, vol. i, c. vi, 8. See vol. ii, pp. 381-5.)
The reasons for the limitation which may have prevented
Michelangelo from adequately representing the sensuous aspect of
womanhood, should be sought in the character of his plastic genius.
So far as the power of appreciation is concerned, and especially in
regard to the spirit of the verse, the opinion of Mr. Symonds appears
to me to reverse the fact. The nature of the artist may be pronounced
especially sensitive to the physical influence of woman. If, in the
extant poetry, this sentiment appears in chastened form, such
calmness may be set down solely to the period of life. Yet even in
these later compositions, extreme impressibility is revealed in every
line. Mr. Symonds’s error has prevented him from entering into the
spirit of the sonnets, and also constitutes a deficiency in his
instructive biography. (See note to sonnet No. 13 [xxx].)
7 [xxii] The verse, direct and passionate, though doubtless of a
later date, still bears the character of pieces which must be
pronounced relatively early. Observable is the use of theologic
metaphor, employed only for the sake of poetic coloring, and not yet
sublimed to pure thought.

8 [xxiv] This delightful sonnet, according to the nephew, was found

on a letter bearing date of 1529. (See p. 7.) The lines seem to give
the idea of a gentle and lovely personage whose countenance
shines out as through a golden mist. In later compositions, the
conflict of Death and Love is worked out differently. (See madrigal
No. 9 [xiii].)
9 [xxv] On the same authority, this inexpressibly charming
production is assigned to 1529. Here appear the germs of Platonic
imagination. The soul, a divine essence, endows the visible
suggestion with the spiritual essence derived from its own store. But
the object is not completely divinized; the end is still possession. The
reflective element will increase, the sensuous lessen, until poetry
passes over into piety.
10 [xxvii] The love verse is not to be taken as wholly biographic,
but rather as ideal.
11 [xxviii] The atmosphere of the sonnet is that of later time and of
a more rarefied height. We are now in full Platonism. The soul,
heaven-born, perceives in the eyes of the beloved its primal home,
the Paradise whence itself has descended, and the heavenly
affection of which earthly love is a reminiscence. But the period may
still be before the Roman residence, and the meeting with Vittoria
Colonna.

12 [xxix] This sonnet may safely be set down as belonging to the

later time. The sentiment of unhappy attachment, impossible desire,
wistful loneliness, breathes through the verse. The piece contains
two mystical but grand lines. Whoever has hoped for an elevation
not given to mortals has wasted his thought in the endeavor to
penetrate the recesses of deity, as seed is lost on the stony ground,
and words spent in the limitless air.
13 [xxx] This gentle and tender poem, of the earlier period,
somewhat similar in sentiment to No. 5 [xix], and obviously from the
heart, is penetrated by the same feeling as that discernible in Nos. 8
and 9 [xxiv and xxv]. Varchi, with his characteristic want of
perception, chose to fancy that it might be addressed to a man, like
the following, said to be composed for Tommaso Cavalieri.

[XXXI]
 A CHE PIÙ DEBB’IO MAI L’INTENSA VOGLIA

What right have I to give my passion vent

In bitter plaint and words of sighing breath,
If Heaven, soon or late, apparelleth
Each living soul in mantle of lament?
Why ere his time, invoke the feet of Death,
When Death will come? Nay, rather let my glance

At last dwell peaceful on his countenance,

Since other good my sorrow vanquisheth.
Yet if no power is mine to shun the blow
I court and seek; what help will be my own,
To interpose ’twixt dolor and delight?
Since prison and defeat allure me so,
It is not strange, if naked and alone,
I remain captive of an armèd knight.

The words cavalier armato are supposed to have referred to the

aforesaid Cavalieri, a Roman youth whom Varchi describes as all
that was beautiful and lovable. The highest male beauty seems to
have had for Italians of the Renaissance, an attraction similar to that
which it possessed for Athenians, a charm which our modern taste
does not entirely comprehend. Thus the early death of Cecchino
Bracchi had produced a great sensation; the epitaphs addressed to
his memory by Michelangelo, who had never looked on his face,
attest the sincerity of his own sentiment. For Cavalieri, whom the
artist had known in 1533, he seems to have what can be described
only as a passion; the three extant letters addressed to the young
man breathe that timidity, sense of inferiority, and fear of
misunderstanding which ordinarily belong only to sexual attachment.
This emotion needs no apology other than that contained in a letter
to this friend: “And if you are sure of my affection, you ought to
think and know that he who loveth remembereth, and can no more
forget the things he fervently loves, than a hungry man the food that
nourishes him; nay, much less may one forget beloved objects than
the food on which man liveth; for they nourish both soul and body,
the last with the greatest sobriety, and the first with tranquil felicity
and the expectation of everlasting salvation.” (Lettere, No. 4, 16.)
The susceptibility of Michelangelo toward external impressions is
noted by Giannotti, who makes him affirm that as often as he set
eyes on any person endowed with excellence he could not help
becoming enamored of him in such manner that he surrendered
himself to him as a prey. (Guasti, Rime, p. xxxi.) To the point is
Michelangelo’s own estimate of his character expressed in a sonnet.

[XVIII]

AL COR DI ZOLFO, ALLA CARNE DI STOPPA

The heart of sulphur and the flesh of tow,

The bones inflammable as tinder dried,
The soul without a bridle, without guide,
In liking prompt, toward joy o’erswift to go,
The reason purblind, halting, lame, and slow,
Tangled in nets wherewith the world doth teem,
No marvel ’tis, if even in a gleam

I kindle up in flame that first doth glow.

With that fair art endowed, whereby the mind
From heaven that bringeth, Nature doth outvie,
And with itself all being occupy,
If I thereto was born nor deaf nor blind,
Proportionate to heat that I desire,
’Tis fault of him who made for me the fire.

It is well to know that Cavalieri seems to have had a modest and

noble nature, and that his personal attachment and artistic
appreciation soothed the declining days of Michelangelo, at whose
end he was present.
 The mention of Michelangelo himself (Lettere, No. 466; Symonds,
Life, vol. ii, p. 130) seems to prove that this sonnet was really
composed for his young friend. But it is one thing to conclude that
the piece was addressed to Cavalieri, quite another to suppose that
it was inspired by him. The ideas are the same as those elsewhere
appearing in reference to women. The composition does not appear
to me one of the most original, and I should be disposed to regard it
as ordinary love verse, into which, out of compliment, the writer had
introduced the punning allusion. In any case, it is to be observed that
in the Platonic compositions treating of male friendship, the whole
argument is metaphorical, the comparisons being borrowed from
the earlier poetry of sexual love.
Fundamental is the question, What proportion of Michelangelo’s
verse was intended to relate to men, and how far can such verse, if
existent, be taken to imply that he had no separate way of feeling for
women? The opinions of Mr. Symonds have already been cited (see
note to No. 6 [xx]). In noticing Michelangelo’s use of the idiomatic
Tuscan word signore, lord, as applied in the sonnets to female
persons as well as male (the English liege may similarly be used), he
says, “But that Michelangelo by the signore always or frequently
meant a woman can be disproved in many ways. I will only adduce
the fragment of one sonnet” (No. lxxxiii). It is a pity that Mr.
Symonds did not enter into detail; I am quite at a loss for any
circumstances that can be held to warrant his declaration. For the
word, the sonnets only afford information. No. xvi, containing the
words signior mie car, is a variant of No. xv, expressly addressed to
a lady. In No. xxii, no one will doubt that the reference is to a
woman. In No. xxxv the sex is shown by the epithet leggiadre, fair,
applied to the arms (Mr. Symonds renders “fragile”). No. xxxvii
qualifies signor by donna. No. lv treats of the shyness of a lady in
presence of her lover. In No. xl, instead of signior, the variant
gives donna. No. xlvii seems obviously addressed to Vittoria
Colonna. In No. xxxvi, the feminine application appears to be
indicated by the description of the sovereign person as reigning nella
casa d’amore. Thus in not a single instance can the suggestion of
Mr. Symonds be accepted.
 There remains the fragment mentioned, No. lxxxiii, a beautiful
and interesting piece, unhappily imperfect. “Yonder it was that Love
(amor; variant, signior), his mercy, took my heart, rather my life; here
with beauteous eyes he promised me aid, and with the same took it
away. Yonder he bound me, here he loosed me; here for myself I
wept, and with infinite grief saw issue from this stone him who took
me from myself, and of me would none.” It will be seen that the
masculine pronoun is rendered necessary by the reference to
personified Love, and that the allusion is clearly to sexual passion.
Mr. Symonds has not entirely comprehended the scope of the
fragment. The mystical description of Love as issuing from a stone
(sasso) may probably be an application of the familiar sculpturesque
metaphor.
As, in the instances considered, the opinion of Mr. Symonds
appears void of foundation, so it is counter to the tenor of the poetry.
If No. xxxi really was written for Cavalieri, the reference probably
consisted of no more than the introduction, into the ordinary phrases
of a love poem, of a complimentary play on words. As for the
metaphor by which a lady is compared to an armed enemy, that was
already commonplace in the day of Dante.
14 [xxxii] From pieces dealing with ideal affection we pass to one
obviously biographic in its inspiration. The poem is written below a
letter of 1532, addressed to the sculptor when in Rome. The artist
seems to refer to his own impetuous nature, too liable to quarrel with
friends. Analogous is the sonnet addressed to Luigi del Riccio. (See
madrigal No. 3 [iv] note.) But this composition evidently relates to a
lady, as is shown by the mention of the dorato strale, gilded dart of
Love.
15 [xxxiii] As with all lyric poetry, so in the compositions of
Michelangelo, it is not to be assumed that every expression of
emotion of necessity corresponds to some particular experience. Yet
the tenderness, melancholy, and gentle regret which inspire the
verse evidently reflect the character and habitual manner of feeling
of the author. Related in sentiment are the following sonnets:—
 [XXVI]

NON MEN GRAN GRAZIA, DONNA, CHE GRAN DOGLIA

By happiness as deep as agony

Below the scaffold is the caitiff slain,
When lost to hope, and ice in every vein,
His pardon comes, his sudden liberty;
So when, beyond thy wonted charity,
My heaven overcast with many a pain
Thy sovereign pity doth make clear again,
More deep than anguish, pierceth ecstasy.
Sweet news and cruel in so far agree,
As either in a moment may destroy
The heart by grief, or sunder it through joy;
If thou desirest that I live for thee,
The rapture mete, for many a creature frail
Hath died of grace too free for its avail.

[XXXV]

SENTO D’UN FOCO UN FREDDO ASPETTO ACCESO

I See a face that in itself is cold,

Yet lit with fire that burneth me afar;
Two arms, that quiet and unmoving are,
Whereby all else is moved and controlled;
The vision of a beauty I behold,
Immortal, yet pursuing me to death;

A power that free, my own envelopeth;

Another’s balm, that may my hurt enfold.
How can befall, that a fair countenance
Hath power to cause effect so contrary,
 Creating what it doth now own? Perchance,
The life that taketh my felicity,
Yet doth itself deny, is like the sun,
That yieldeth the world heat, yet heat hath none.

[XXXVIII]

RENDETE A GLI OCCHI MIEI, O FONTE O FIUME

Ah give me back, or river thou or source,

The turbid waters that enlarge thee so,
That thy augmented current doth o’erflow,
And hasten on an unaccustomed course;
O laden air, whose gathered mists allay
And temper heaven’s shining to these eyes,
Return my weary heart her many sighs,
And cloudless leave the countenance of day.
Earth, render to my feet their steps again,
Along the track they trod let grass grow green;
Restore, deaf Echo, my petitions vain;
And ye, alas! unmovèd eyes serene!
Give mine their wasted looks, that they may see
Some kinder loveliness, disdained by thee!

With these sonnets of ideal love may be compared one later in

date, apparently more biographic in sentiment, and doubtless
inspired by Vittoria Colonna.

[L]

S’I’ AVESSI CREDUTO AL PRIMO SGUARDO

 Had I believed, when first I met the glow
Of this bright soul, my sun, that I might rise
Through fire renewed in such triumphal wise
As doth the Phœnix from her ashes go,
Like some fleet-footed creature, pard or roe,
That seeks its joy and flieth from its fear,
To meet the act, the smile, the accent dear,
I would have leaped, now in my swiftness slow.
Yet why indulge regret, the while I see
In eyes of this glad angel, without cease,
My calm repose and everlasting peace?
More painful days, perchance, had dawned on me,
If I had earlier met, yet been denied
The wings she lendeth me to fly beside.

16 [xxxix] The timid lover, who finds himself involved in the

dangers of a hopeless passion, endeavors to withdraw from the
perilous situation, but in so doing finds himself confronted by another
danger, that of losing the affection which has become his life. As the
vain desire will prove the death of the body, so the renunciation will
be that of the soul; thus the suitor, according to the familiar
metaphorical system furnished by plastic art, is said to see his
lady with a statue of Death on either hand.
The beautiful and mystic sonnet was written on a stray leaf
bearing a memorandum of 1529, and was probably composed in that
year. According to the statement of the nephew, Nos. 8 and 9 [xxiv
and xxv] were also written on letters of that year; and these two
poems correspond in sentiment with the present piece.
17 [xl] This most beautiful sonnet might conjecturally be referred
to the same period as No. 12 [xxix]. The spirit of the verse ought to
be enough to satisfy any reader that it was composed with reference
to a woman. (See note to No. 13 [xxx].)
18, 19 [xliii, xliv] These two pieces, containing respectively the
dispraise and praise of night, are obviously intended to be
counterparts, the first forming an introduction to the second. The
consolations belonging to darkness and slumber have furnished
themes to very many writers of verse; but among all such pieces
Michelangelo’s tribute is entitled to preëminence. The emotion,
deepening with the progress of the rhyme, ends in one of those
outbursts which make the poetry a key to the character. Two other
sonnets treating of the same subject do not appear to be connected.

[XLI]

COLUI CHE FECE, E NON DI COSA ALCUNA

He who did erst from primal nothing bring

Time, integral and property of none,
To half, dividing, gave the distant sun,
To half the moon, a lamp more neighboring.
All in a moment, Destiny and Chance
Began, and over mortals ruled with power;
To me they gave the still and sober hour,
As like to like, in birth and circumstance.
As attribute in action is expressed,
And darkness is the property of night,
So e’en to be myself is sad to be;
Yet is my troubled spirit soothed to rest,
Remembering, its dusk may render bright
The sun that Fortune lent for friend to thee.

In No. xlii Night is lauded, as the shadow in which man is

engendered, while in the day the soil is broken only for the seed of
the corn; but the composition does not rival the sweetness and
sublimity of No. xliv.
20 [lii] This fine sonnet, belonging to the later period, may be set
down as among those inspired by Vittoria Colonna. Thoroughly
characteristic is the grand fifth line, in which the soul is said to have
been created as God’s equal. The nephew, of course, diluted such
daring conceptions into commonplace, and his restoration
altogether fails to convey the essential meaning of the piece.
Wordsworth, unfortunately, knew only the emasculated version.
Similar in theme is another sonnet, No. lx, also rendered by
Wordsworth, from a text more nearly representative. In this instance
the English poet has transcended his source, and furnished a proof
that on fortunate occasions a translation may belong to the very best
poetry, and deserve that immortality which commonly belongs only to
expressions of original genius.

Yes! hope may with my strong desire keep pace,

And I be undeluded, undismayed;
For if of our affections none find grace
In sight of heaven, then wherefore hath God made
The world which we inhabit? Better plea
Love cannot have, than that in loving thee
Glory to that eternal peace is paid,
Which such divinity to thee imparts,
As hallows and makes pure all gentle hearts.
His hope is treacherous only, whose love dies
With beauty, which is varying every hour;
But in chaste hearts uninfluenced by the power
Of earthly change, there blooms a deathless flower,
That breathes on earth the air of Paradise.

21 [lvi] The sonnet is to be classed with the preceding. In a

variant, the theologic metaphor is carried further: “From without, I
know not whence, came that immortal part which separateth not
from thy sacred breast, yet traverseth the entire world, healeth every
intellect, and honoreth heaven.”
22 [lxi] As all tools used by man are formed by means of other
tools, the archetypal tool must be that celestial instrument by which
the world is fashioned. On earth, Vittoria Colonna had been the
hammer (as we now say, the chisel) by which had been inspired the
creative activity of the artist. By her death, this influence had been
withdrawn to heaven, there to become united with the all-forming
hammer of the eternal Maker; it is, therefore, only from on high that
the artist can look for the completion of his own genius.
To the text, in the hand of Michelangelo, is added a sentence
expressing his sense of the incomparable merit of Vittoria, as the
divine instrument which none other is able to wield, and a prayer that
his own hammer, as he metaphorically says, may also attain a
reception in heaven.
The mystically expressed, but in reality simple and direct verse is
crowded with ideas which strive for utterance. The sculptor
seems to have written prophetically; after the passing away of
Vittoria, the last of his animating impulses appears to have been
removed, and his life becomes that of a recluse, struggling with the
infirmities of advancing age.
Several other pieces relate to the death of Vittoria.

[LXII]

QUAND’EL MINISTRO DE’ SOSPIR ME’ TANTI

When she who ministereth sighs, withdrew

From the world’s sight, from her own self and mine,
Nature, who made her in our eyes to shine,
Remained abashed, and downcast all who knew.
Yet be not Death of his loud vauntings rife
O’er the sun’s sun, as over others; Love
Hath him subdued, and her endowed with life
Both here below, and with the pure above.
Unjust and haughty Death did so engage
To hush her praises, and her soul bestow
Where it would seem less beautiful; and lo!
Reverse effects illuminate Time’s page;
On earth, more life than she in life possessed,
 While Heaven who wished her, now enjoyeth blest.

The thought, that Nature is disgraced in the loss of its best

creation, is repeated in Michelangelo’s poetry. (See sonnet No. 4
[xvii], madrigal No. 9 [xiii].)

Two other sonnets, Nos. lxiii and lxiv, breathe an atmosphere of

the most gloomy despair. The first expresses a profound self-
reproach; the time to soar heavenward was while the sun of life still
shone; it is now too late. The second declares that the flame has
expired, to leave only ashes without a spark.
I do not doubt that here also belongs another sonnet, placed by
Guasti as if belonging to an earlier date.

[LI]

TORNAMI AL TEMPO ALLOR CHE LENTA E SCIOLTA

Give me the day when free was cast the rein

For headlong ardor’s unreflecting race;
Restore to me the calm angelic face
Wherewith interred seems Virtue to remain;
Give back the wanderings, the steps of pain,
So slow to him by weary age oppressed;
Give water to my eyes, fire to my breast,
If thou wilt take thy fill of me again.
If, Love, ’tis true, thou livest on no more
Than sighs and tears of lovers bitter-sweet,
A weary age hath nought of thy desire;
The soul already near the further shore,
With shield of holier darts doth thine defeat,
And the burned wood is proof against the fire.
 A madrigal relates to the same theme.

[VI]

PER NON S’AVERE A RIPIGLIAR DA TANTI

That perfectness of beauty free from peer

Might be reclaimed and garnered without fail,
Upon a lady excellently clear
Was it bestowed beneath a shining veil;
The heavenly boon had hardly been repaid,
If scattered among all that Heaven had made.
Now, from world unaware,
In breathing of a sigh,
Hath God who reigns on high
Resumed, and hid away his beauty fair.
Yet, though the body die,
Cherish shall memory still
Sacred and sweet, her written legacy.
Compassionate and stern, if Heaven’s will
To all had granted what to her alone,
We all had died for making good the loan.

The madrigal recites that deity had chosen to embody in a single

life the sum of beauty, to the end that the celestial gift might be more
easily resumed. Similar concetti are to be found in the series of
epitaphs composed on Cecchino Bracci, in 1544. Mr. Symonds very
unjustly criticises the verse as constrained, affected, and exhibiting
an absence of genuine grief.
 NOTES ON THE EPIGRAMS

1 [i] the night of the medici chapel. According to Vasari, when

the statues of the Medici Chapel were exposed to view, after
Michelangelo’s departure for Rome, early in 1535, an unknown
author affixed a quatrain to the image of Night. This person was
afterwards known as Giovanni di Carlo Strozzi, at the time eighteen
years of age. The verse, not ungraceful but superficial, recited that
Night, carved by an angel, was living, for the very reason that she
seemed to sleep, and if accosted, would make reply. To this fanciful
compliment, Michelangelo responded in the beautiful quatrain, which
exhibits his view of the Medicean usurpation.
It were to be wished that in presence of the awful forms, visitors
would bear in mind the sculptor’s advice. I have heard a young
American lady, in a voice somewhat strident, expound to her mother
the theme of the statue, reading aloud the information furnished by
Baedeker.
2 [ii] death and the coffin. The younger Buonarroti cites the
statement of Bernardo Buontalenti, that in his house in Rome,
halfway up the stair, Michelangelo had drawn a skeleton Death
carrying on his shoulder a coffin, on which were inscribed these
lines. The story is interesting, in connection with the part taken by
Death in the verse of the sculptor. Giannotti represents him as
declining to attend a merry-making on the ground that it was
necessary to muse on Death. (See madrigal No. 12 [xvi].) The idea
appears to be that death cannot be dreadful, since it bequeaths to
life not only the immortal soul, but even the body; probably the artist
meant to say the body made immortal through art.
3 [v] definition of love. With this definition from the subjective
point of view, may be compared madrigal No. 5 [viii]. As usual the
imagination of the poet takes plastic form; Love, in his mind, is a
statue lying in the heart, and waiting to be unveiled. Akin is the
celebrated sonnet of Dante, Amor e cor gentil sono una cosa, which
contains the same conception, and which perhaps Michelangelo may
have remembered. But the more mystical idea of the sculptor
borrows only the suggestion.
 NOTES ON THE MADRIGALS

1 [i] During his Roman residence, Michelangelo was brought into

intimate relations with Florentine exiles, who gathered in Rome,
where ruled a Farnese pope, and where certain cardinals favored
the anti-Medicean faction. From the course of a turbulent mountain-
brook, Florence, following an inevitable law, was obliged to issue into
the quiet but lifeless flow of inevitable despotism. It could not be
expected that the fiery Michelangelo could comprehend the
inexorableness of the fate which, in consequence of the necessities
of trade, compelled Florence to prefer conditions ensuring
tranquillity, though under an inglorious and corrupt personal rule. The
sublime madrigal shows the depth of his republican sentiments. (See
No. 22 [lxviii].)
2 [iii] The difficult but very interesting madrigal gives a profound
insight into the spirit of the writer, who felt himself to move in a
society foreign from the higher flight of his genius. His habits of
isolation are remarked by contemporaries. Giannotti, in the dialogue
above mentioned, discourses amusingly on this trait of character,
putting into the mouth of the artist a reply to an invitation. “I won’t
promise.” “Why?” “Because I had rather stay at home.” “For what
reason?” “Because, if I should put myself under such conditions, I
should be too gay; and I don’t want to be gay.” Luigi del Riccio,
introduced as interlocutor, exclaims that he never heard of such a
thing; in this sad world one must seize every opportunity of
distraction; he himself would supply a monochord, and they would all
dance, to drive away sorrow. To this comforting proposition,
Michelangelo returns that he should much prefer to cry. Giannotti
romances; but Francis of Holland is nearer the fact when he makes
the sculptor answer an accusation urged against solitary habits. The
artist declares that there is good ground for such accusation against
one who withdraws from the world by reason of eccentricity, but not
against a man who has something better to do with his time. The
particular occasion of the madrigal seems to have been
dissatisfaction with praise lavished on what to Michelangelo seemed
an unworthy work. Southey paraphrases the poem, but gives the
idea only imperfectly.
Here, in connection with the idea of beauty as furnished from
within, may be introduced a version of a madrigal interesting rather
on account of the philosophic conception than the poetic
excellence. (See also sonnet xviii, translated in the note to No. xxx.)

[VII]

PER FIDO ESEMPLO ALLA MIA VOCAZIONE

On me hath been bestowed by birthtide-gift,

Of both mine arts the mirror and the light,
Beauty, my model in my calling here.
It only hath the competence to lift
The vision of the artist to that height
At which I aim in form or color clear.

If judgment rash and fantasy unwise

Degrade to sense the beauty, that doth bear
And raise toward heaven all sane intelligence,
Man’s wavering glances have no power to rise,
Above inconstant, faithful only where
They linger, unless mercy call them thence.

3 [iv] The madrigal is addressed to Luigi del Riccio, friend of

Michelangelo’s declining years, and a correspondent to whom were
transmitted many of the extant poems. In 1544 Luigi, during a
sickness of the sculptor, took him into his own house and acted as
his nurse; but shortly afterwards, he refused a request of the artist,
declining to suppress an engraving he had been requested to
destroy. The indignation of Michelangelo found vent in a bitter
letter. Riccio died in 1546. Symonds (Life, vol. ii, p. 194) thinks that
Michelangelo speedily excused his friend and repented his anger.
Here the whole heart of the artist is disclosed, and we have a
revelation of the manner in which internal brooding and many
disappointments had rendered somewhat morose a gentle and
affectionate nature, characterized by pride amounting to a fault.
With the idea may be compared Emerson’s essay on “Gifts.”
“Hence the fitness of beautiful, not useful things, for gifts. This giving
is usurpation, and therefore, when the beneficiary is ungrateful, as all
beneficiaries hate all Timons, not at all considering the value of the
gift, but looking back to the greater store it was taken from, I rather
sympathize with the beneficiary than with the anger of my lord
Timon. For the expectation of gratitude is mean, and is continually
punished by the total insensibility of the obliged person. It is great
happiness to get off without injury and heart-burning from one who
has had the ill-luck to be served by you. It is a very onerous
business, this of being served, and the debtor naturally wishes to
give you a slap.” He adds, entirely in the spirit of Michelangelo, “No
services are of any value, but only likeness.”

4 [v] The poet addresses to his friend Vittoria Colonna a theologic

inquiry, after the manner of the appeals of Dante to Beatrice.
Apparently the letter included a blank leaf for an answer. The
question is, “In heaven are contrite sinners less valued than self-
satisfied saints?” The obvious reply must be that in the nature of
things such saints are impossible. The inquiry, therefore, is not to be
taken as serious, but as playful and ironical. I should be inclined to
interpret the verse as asking, “Am I, an humble artist, but sincerely
devoted, of less value in your eyes than the very courtly and
important personages by whom you are surrounded?” (as Vittoria
was in close intimacy with high ecclesiastical functionaries). The
sentiment is gay and jesting, while full of pleading affection.
5 [viii] If of all the compositions of Michelangelo, one were asked
to name the most representative, it would be natural to select this
incomparably lovely madrigal. No lyric poet has brought into a few
words more music, more truth, more illumination. The four lines cited
at the end of the Introduction might well be taken as the motto for a
gathering of the poems; and if the arrangement had not seemed
inconsistent with the numbering of the pieces, I would gladly have
placed the madrigal at the end, as summing up the especial
contribution of Michelangelo to letters.
6 [ix] A charming and light-hearted piece of music, obviously
belonging to the earlier period of Michelangelo’s poetic activity. The
verse is written on blue paper, with the subscription, “Divine things
are spoken of in an azure field” (in heaven). The suggestion is
furnished by a conventional concetto of the period; but the familiarity
does not prevent the thought lending itself to genuinely poetical
treatment. No. x is a pretty variant, in which the cruelty of the lady is
compared to the hardness of the marble in which her image is
wrought. The lines are subscribed “for sculptors” (Da scultori). The
close connection with his art lends to even the most simple of these
verses an unspeakable attraction.
7 [xi] In this magnificent song, worthy of the greatest of lyric poets,
we are still occupied with the concepts of plastic art. The artist
achieves the complete expression of his idea only through painful
toil, and often lapse of years which leave him ready to depart from a
world in which accomplishment is itself a sign of ripeness for death.
With that universal animism, as we now say, by which all general
truths of man’s life are felt to be also applicable to the course of
Nature, the poet is entitled to apply the idea to external being.
And with what insight! If ever genius can be said to have forecast the
conclusions of scientific inquiry, it is so in this instance; Michelangelo
presents us with a truly modern conception of Nature, as the creative
artist, who through a series of ages and a succession of sketches, is
occupied with continually unsuccessful, but ever-improving efforts at
the expression of her internal life. The perfection of the creature,
which marks the accomplishment of the undertaking, signifies also
the end of the process; with such completeness is felt the sorrow
incident to all termination, and especially the pain of the mortal, who
feels that delight in perfect beauty enforces the consciousness of his
own transitoriness, and emphasizes the sense of Nature as